A mechanical arm path planning method and system based on an improved escape optimization algorithm

Abstract

This invention provides a robotic arm path planning method and system based on an improved escape optimization algorithm, belonging to the field of path planning technology. Specifically, it includes: a model building module responsible for acquiring a working environment map and performing 3D modeling to obtain an environment map model and boundary constraints; a population generation module responsible for merging the original path population with the new path population and selecting the optimal one as the initial path set; an iterative optimization module responsible for performing collision detection and fitness calculation on the smooth trajectories generated by B-spline interpolation of candidate paths, and updating the globally optimal path based on the calculation results; and a trajectory generation module responsible for converting the globally optimal path into a robotic arm joint space path, performing B-spline interpolation optimization again, and generating the final smooth joint trajectory driving the robotic arm movement, thus improving the efficiency of the driving process.

Description

Technical Field

[0001] This invention belongs to the field of path planning technology, and in particular relates to a path planning method and system for robotic arms based on an improved escape optimization algorithm. Background Technology

[0002] With the rapid development of modern industrial automation, intelligent logistics, and special operations, robots, especially multi-degree-of-freedom robotic arms, are playing an increasingly important role in many application scenarios. Their efficient, safe, and autonomous operation capabilities rely heavily on one of their core technologies—path planning. Path planning aims to find an executable trajectory for the robot from a starting point to a destination within a given working environment. This path must not only meet collision-free constraints but also consider various performance indicators such as path length, time, smoothness, and energy consumption, especially in complex scenarios with complex obstacles, narrow passages, or rugged terrain.

[0003] Currently, various methods have been developed in the field of robot path planning, which can be mainly categorized as follows: Graph search-based methods, such as A* and Dijkstra's algorithm, optimize through a discretized search space. Random sampling-based methods, such as PRM, RRT, and their variants (RRT*), construct feasible paths through random exploration. Biomimetic intelligent optimization algorithms, such as ACO, PSO, and GA, iteratively optimize by simulating natural behavior, demonstrating strong global search capabilities. However, existing intelligent optimization algorithms still face the following key challenges when applied to path planning in complex scenarios for robotic arms: traditional random initialization easily generates a large number of invalid solutions; population diversity decreases in the later stages of the algorithm, making it prone to getting trapped in local optima; and it is difficult to achieve a good balance between convergence speed and accuracy.

[0004] In summary, to solve the above-mentioned technical problems, this application provides a robotic arm path planning method and system based on an improved escape optimization algorithm. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a robotic arm path planning system based on an improved escape optimization algorithm, which includes: The model building module is responsible for acquiring the working environment map and performing 3D modeling to obtain the environment map model and boundary constraints. The population generation module is responsible for generating the original path population based on the boundary constraints using random initialization, and performing adaptive Gaussian perturbation on a portion of the selected original path population to generate a new path population. The original path population and the new path population are merged and optimized to form the initial path set. The iterative optimization module is responsible for updating the paths in the initial path set into calm groups, panic groups, and conformist groups based on the fitness of the paths. Specifically, the calm and panic groups are updated based on a spiral search strategy combined with a dynamic panic index, while the conformist group is updated based on the attraction to the center of the calm group and the global optimum. After the number of iterations reaches the target number of iterations, a dynamic mutation mechanism is introduced to improve the ability of candidate paths to escape local optima. A multi-objective fitness function is constructed, which includes path length, collision penalty, path height, and path smoothness. Collision detection and fitness calculation are performed on the smooth trajectories generated by B-spline interpolation of candidate paths, and the global optimum path is updated based on the calculation results. The trajectory generation module is responsible for converting the global optimal path into a joint space path of the robotic arm, and then performing B-spline interpolation optimization to generate a smooth joint trajectory that ultimately drives the robotic arm to move.

[0006] Furthermore, the boundary constraints are constructed using the space-occupying markers of static obstacles in the work environment map.

[0007] Furthermore, the original path population and the new path population are merged and optimized to form the initial path set, specifically including: Based on the preset population size and search space, and in conjunction with the boundary constraints, the original path population is generated using random initialization. From the original path population, a preset number of original valid paths are randomly selected as base paths, and a new path population is generated by performing multiple Gaussian perturbations on each selected base path. The original path population and the new path population are merged, their fitness is calculated and sorted from low to high, and the top-ranked populations are selected as the final initial paths.

[0008] Furthermore, based on the fitness of the paths, the paths in the initial path set are divided into calm group, panic group, and conformity group for updating, specifically including: The paths in the initial path set are sorted in ascending order of fitness, and the paths are divided into calm group, panic group and conformity group according to a preset ratio; The paths of the calm group and the panic group are updated based on the spiral search strategy of the whale optimization algorithm, and the panic index is introduced as a random perturbation term. The path of the conformist group is updated based on its own inertia, the center of the calm group, and the attraction of the global optimal solution, and a boundary adaptive exploration mechanism is introduced based on the boundary gravity.

[0009] Furthermore, the dynamic mutation mechanism includes uniform spherical mutation and jump mutation.

[0010] Furthermore, the radius of the uniform and jump variations of the sphere is dynamically adjusted based on the panic index and the diagonal length of the map size.

[0011] Secondly, the present invention provides a robotic arm path planning method based on an improved escape optimization algorithm, applied to the aforementioned robotic arm path planning system based on an improved escape optimization algorithm, specifically including: S1 determines the iterative processing time of the robotic arm in the path planning process based on the path planning data, and determines the comparison and analysis processing time range of the image features of the working environment map based on the iterative processing time of different path iterative processing processes. S2 determines the acquisition and processing scheme of the identified image features by utilizing the similarity of image features within different comparison analysis time intervals; S3, based on the acquisition and processing scheme, performs the update processing of the recognition image features. According to the analysis results of the similarity of the recognition image features, the recognition image features are divided into different groups. Based on the recognition image features in the group and the deviation of the iteration processing time between the recognition image features, the target iteration number adjustment processing strategy in the group is determined. S4 adjusts the target iteration number in the group according to the adjustment processing strategy, and determines the target iteration number identification processing method according to the adjustment processing strategy of the target iteration number in different groups and the adjustment data in different iteration number intervals.

[0012] Furthermore, the path planning data of the robotic arm includes the number of path planning processes of the robotic arm and the iteration processing time of the path planning process.

[0013] Furthermore, the method for determining the time interval for the comparative analysis of image features of the working environment map is as follows: S11 determines the proportion of path iteration processes within different iteration processing time intervals based on the iteration processing time of different path iteration processing processes; S12 determines the process proportion of the iterative processing time interval based on the proportion of the number of path iterative processing processes in different iterative processing time intervals; S13 determines the comparison and analysis processing time interval of the image features of the working environment map based on the process proportion in different iteration processing time intervals.

[0014] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0015] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0016] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0017] Figure 1 This is a framework diagram of a robotic arm path planning system based on an improved escape optimization algorithm; Figure 2 This is a flowchart of merging the original path population with the new path population and selecting the best one to form the initial path set; Figure 3 It is a flowchart that updates the initial path set by dividing the paths into calm groups, panic groups, and conformity groups based on the fitness of the paths. Figure 4 This is a comparison chart showing the adaptability of the DOA method, FTTA method, IRIME method, ESC method, IRRT method and the robotic arm path planning method of this invention in two environments. Figure 5 These are iterative curves of the DOA method, FTTA method, IRIME method, ESC method, IRRT method, and the robotic arm path planning method of this invention in two environments. Figure 6 This is a box plot comparing the DOA method, FTTA method, IRIME method, ESC method, IRRT method and the robotic arm path planning method of this invention in two environments.

[0018] Figure 7 This is a flowchart of a robotic arm path planning method based on an improved escape optimization algorithm. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0020] Example 1 like Figure 1 As shown, this application provides a robotic arm path planning system based on an improved escape optimization algorithm, specifically including: The model building module is responsible for acquiring the working environment map and performing 3D modeling to obtain the environment map model and boundary constraints. The population generation module is responsible for generating the original path population based on the boundary constraints using random initialization, and performing adaptive Gaussian perturbation on a portion of the selected original path population to generate a new path population. The original path population and the new path population are merged and optimized to form the initial path set. The iterative optimization module is responsible for updating the paths in the initial path set into calm groups, panic groups, and conformist groups based on the fitness of the paths. Specifically, the calm and panic groups are updated based on a spiral search strategy combined with a dynamic panic index, while the conformist group is updated based on the attraction to the center of the calm group and the global optimum. After the number of iterations reaches the target number of iterations, a dynamic mutation mechanism is introduced to improve the ability of candidate paths to escape local optima. A multi-objective fitness function is constructed, which includes path length, collision penalty, path height, and path smoothness. Collision detection and fitness calculation are performed on the smooth trajectories generated by B-spline interpolation of candidate paths, and the global optimum path is updated based on the calculation results. The trajectory generation module is responsible for converting the global optimal path into a joint space path of the robotic arm, and then performing B-spline interpolation optimization to generate a smooth joint trajectory that ultimately drives the robotic arm to move.

[0021] Furthermore, the boundary constraints are constructed using the space-occupying markers of static obstacles in the work environment map.

[0022] Furthermore, such as Figure 2 As shown, the initial path set is obtained by merging the original path population with the new path population and selecting the best one. Specifically, this includes: Based on the preset population size and search space, and in conjunction with the boundary constraints, the original path population is generated using random initialization. From the original path population, a preset number of original valid paths are randomly selected as base paths, and a new path population is generated by performing multiple Gaussian perturbations on each selected base path. The original path population and the new path population are merged, their fitness is calculated and sorted from low to high, and the top-ranked populations are selected as the final initial paths.

[0023] The above steps specifically include: Original path population The formula is as follows:

[0024] in, and Representing the first Upper and lower bounds in each dimension ensure that the original path population does not exceed the effective working space. Random variables The uniform distribution between 0 and 1 reflects the randomness of the initial evacuation decision-making process.

[0025] population It is random from choose Quantity The new path population generated by adaptive Gaussian perturbation of a certain number of iterations is usually , New Pathway Population The formula is as follows:

[0026] in, It is random from choose The original population formed by the quantity; For adaptive Gaussian perturbation, the update formula is as follows:

[0027] in, The strength of the Gaussian perturbation based on this; The elements therein follow a standard normal distribution; , , It is a constant with a value of 0.5; The formula is as follows:

[0028] Furthermore, the calculation formula for the multi-objective fitness function is as follows:

[0029] in, Represents fitness. Cost representing path length This represents the cost associated with the height standard deviation. Cost representing path smoothness; Represents the weighting coefficient; This represents a collision penalty.

[0030] Specifically, the fitness of each population group is calculated. The formula is as follows:

[0031] in, Represents fitness. Cost representing path length This represents the cost associated with the height standard deviation. Cost representing path smoothness; This represents the weighting coefficient, which, here, is taken as... , , ; The collision penalty is represented by the following formula:

[0032] Optimizing path length is typically crucial for saving time and reducing costs while ensuring safety. Path length The update formula is as follows:

[0033] The first in the representative path planning The location of each point.

[0034] In addition, the altitude of path planning has a significant impact on the control system and safety, and this factor needs to be considered. The update formula is as follows:

[0035] Finally, the impact of path smoothness must also be considered. The update formula is as follows:

[0036] in represent .

[0037] Furthermore, such as Figure 3 As shown, the paths in the initial path set are divided into calm, panic, and conformity groups based on their fitness for updating. Specifically, this includes: The paths in the initial path set are sorted in ascending order of fitness, and the paths are divided into calm group, panic group and conformity group according to a preset ratio; The paths of the calm group and the panic group are updated based on the spiral search strategy of the whale optimization algorithm, and the panic index is introduced as a random perturbation term. The path of the conformist group is updated based on its own inertia, the center of the calm group, and the attraction of the global optimal solution, and a boundary adaptive exploration mechanism is introduced based on the boundary gravity.

[0038] Specifically, in each iteration Initially, the fear index The calculation is as follows:

[0039] in, As the central reference value, The amplitude coefficient, For curvature parameters, The center value.

[0040] Set the number of elite pools The algorithm begins its iterative process, when the number of iterations reaches... At that time, they were divided into calm Conformity and panic The three categories are updated separately. Specifically, the population is sorted in ascending order of fitness, and the paths are divided into three groups according to proportion: the calm group, the conformist group, and the panic group. .

[0041] Rational behavior within a calm group, moving towards the center. Movement, representing a collective decision-making process within a group, is represented by the following formula:

[0042]

[0043]

[0044] (Mean of the calm path). Represents the first path in the calm group. The minimum and maximum values ​​of the dimension. It indicates a slight adjustment to one's personal movements. The value is 0 or 1 (with equal probability), determined by the Bernoulli distribution, allowing for partial updates. It simulates the partial dimensions that are not updated due to crowding. For the Lévy weights, the updated formula is as follows:

[0045] It is a Gamma function. The parameters will be dynamically adjusted as the algorithm progresses, as shown in the following formula:

[0046] yes The initial value is 1.5, which is the same empirical setting used in the Harris Hawks Optimization (HHO) algorithm. This adjustment allows the algorithm to make larger exploratory moves initially (when...). When the movements are smaller, then gradually transition to more refined movements (when...). (When it increases), it reflects the natural process within a population moving from panic-driven exploration to calm, rational decision-making.

[0047] The conformity path follows the behavior of both the calming and panicking groups. Their formula is updated based on the influence of both:

[0048] in, A set of solutions randomly selected from the panicked crowd represents the potential direction of panic-driven movement. Is it an alternative adaptive Levy weight used for the eligible group? It is a position randomly generated within the range of the herd. Represents the first path among all paths in the group. The minimum and maximum values ​​of the dimension. They are binary variables generated through the same mechanism. The updated formula is as follows:

[0049] Benefiting from potential exports (elite pool) The panic group update formula is as follows, taking into account the influence of other random path indicators:

[0050]

[0051] It is from the elite pool A randomly selected set of solutions represents the possible exits. This represents a path randomly selected from the crowd, introducing an element of randomness into panic-driven movements. It is a location randomly generated within the panic group. This represents the first path in all paths within the panic group. The minimum and maximum values ​​of the dimension.

[0052] When the number of iterations At that time, the calm group, the conformity group, and the panic group were adjusted to... .

[0053] Joint update of the calm and panic groups: For each path in both the calm and panic groups, the update behavior is influenced by a spiral search mechanism that combines whale optimization algorithm with random perturbations of the panic index. When using either a shrinking encirclement strategy or a random search strategy, the formulas are as follows:

[0054] in and The weighting coefficients are updated using the following formula:

[0055] when When using a spiral search strategy, the formula is as follows:

[0056] in To control the constant parameters of the spiral shape, To distribute evenly in A random number between [a certain number of points].

[0057] For each path in the conformity group, its update strategy incorporates its own inertia, the attraction to the center of the calm group and the global optimum, and adaptive boundary exploration. The formula is as follows:

[0058] To prevent the path from stalling near the search space boundary, a random exploration term is added when the path approaches the boundary. The formula is as follows:

[0059] in, The boundary disturbance intensity coefficient, The threshold is the value close to the boundary.

[0060] When the number of iterations is less than or equal to the target number of iterations, for example At this point, all paths are considered calm, and paths optimize their position by moving closer to members of an elite pool, which represents potential safe exits and the best solution determined in previous iterations, as well as paths randomly selected from the population. The formula is as follows:

[0061] The position for a member of the elite pool symbolizes a possible safe exit and one of the best solutions identified to date. It is the location of a path randomly selected from the population.

[0062] Furthermore, the dynamic mutation mechanism includes uniform spherical mutation and jump mutation.

[0063] Furthermore, the radius of the uniform and jump variations of the sphere is dynamically adjusted based on the panic index and the diagonal length of the map size.

[0064] Furthermore, the radius of the uniform and jump variations of the sphere is dynamically adjusted based on the panic index and the diagonal length of the map size.

[0065] Specifically, when the number of iterations is less than or equal to the target number of iterations, that is... At this point, all paths are considered calm, and paths optimize their position by moving closer to members of an elite pool, which represents potential safe exits and the best solution determined in previous iterations, as well as paths randomly selected from the population. The formula is as follows:

[0066] The position for a member of the elite pool symbolizes a possible safe exit and one of the best solutions identified to date. It is the location of a path randomly selected from the population.

[0067] When the number of iterations At that time, an adaptive breakthrough strategy for survival in dire situations is adopted.

[0068] The radii of the uniform and jump variations of the sphere are dynamically adjusted based on the panic index and the diagonal length of the map size. The formula is as follows:

[0069] A random number between [0,1] If a uniform spherical mutation (local search) is performed, then a jump mutation (global search) is performed. Generation direction. The search radii for both uniform and jump mutations of the sphere are dynamically adjusted using formulas. and formula Update. If the solution is located at the map edge, it can easily generate many invalid solutions, so a reverse correction matrix is ​​constructed, as shown in the following formula:

[0070] in , Therefore, the search radius for uniform variation of the sphere is ,in The search radius for jump mutation is... ,in . For map dimensions, the formula is as follows:

[0071] If the mutated solution exceeds the boundary, it is processed by reflection using the following formula:

[0072] Step 4: Calculate the fitness F in each iteration and record the optimal solution for each iteration. With optimal fitness, .

[0073] The generated path G_sol is interpolated using a fifth-order B-spline optimization method to obtain the planned path of the robotic arm.

[0074] Furthermore, the specific method for performing collision detection and fitness calculation on the smoothed trajectories generated by B-spline interpolation of candidate paths, and updating the globally optimal path based on the calculation results, is as follows: After generating a smooth trajectory by B-spline interpolation for each candidate path, collision detection and fitness calculation are performed sequentially. The fitness calculation result is compared with the fitness value of the current global optimal path. If the fitness value of the current candidate path is better than that of the global optimal path, the global optimal path is updated with the current candidate path.

[0075] (1) Collision detection processing: The collision detection process refers to the process of determining whether the body geometric envelope of the robotic arm in that pose interferes with the geometric model of the obstacle by judging the discrete sampling points on the smooth trajectory generated by B-spline interpolation. Specifically, the method is as follows: First, the smooth trajectory is uniformly and discretely sampled at a preset sampling interval to obtain a set of sampled pose sequences; then, for each sampled pose, the position and orientation of each link of the robotic arm in Cartesian space are calculated through forward kinematics; then, using the bounding box model or capsule model of each link as the geometrically simplified representation of the robotic arm, the minimum distance between the bounding box of each link and the established pixel map or point cloud model of the obstacle in the workspace is calculated; if the minimum distance between any link and any obstacle is less than a preset safe distance threshold, it is determined that there is a collision risk at that sampling point, the candidate path is marked as a collision path, and a collision penalty term is calculated based on the degree of collision; if there is no collision risk at any sampling point, the candidate path is determined to be a collision-free path.

[0076] The collision penalty term is calculated as follows: count the number of sampling points on the smooth trajectory whose minimum distance is less than a preset safety distance threshold, and use the ratio of this number to the total number of sampling points as the collision ratio. The product of the collision ratio and the preset collision penalty coefficient is used as the collision penalty term for the candidate path. The higher the collision ratio, the larger the collision penalty term.

[0077] (2) Fitness calculation process: The fitness calculation process refers to the process of comprehensively evaluating the merits of candidate paths across multiple dimensions, including path quality, motion efficiency, and safety, to generate a unified fitness value. The fitness function is a weighted sum of path length cost, trajectory smoothing cost, joint variation cost, and collision penalty cost, and its specific expression is: F = w1·L + w2·S + w3·J + w4·C in: L is the path length cost term, which refers to the total arc length of the smooth trajectory in Cartesian space, reflecting the movement distance of the path. The shorter the arc length, the lower the path cost. S is the trajectory smoothing cost term, which refers to the sum of the absolute values ​​of curvature at each sampling point on the smoothed trajectory. It reflects the degree of geometric smoothness of the trajectory. The smaller the sum of curvature, the smoother the trajectory. J is the joint change cost term, which refers to the sum of the absolute values ​​of the joint angle changes of all joints in the corresponding joint space path between adjacent path points. It reflects the amplitude of joint movement. The smaller the joint change, the more continuous the movement. C is the collision penalty term, which is calculated by the collision detection process; w1, w2, w3, and w4 are the preset weight coefficients for the four cost items mentioned above. Each weight coefficient is set according to the actual application scenario, focusing on the path length, smoothness, joint motion, and safety. The sum of the four weight coefficients is 1.

[0078] The smaller the fitness value, the better the overall quality of the candidate path.

[0079] (3) Global optimal path update processing: The global optimal path update process refers to comparing the fitness value of the current candidate path with the historical best fitness value recorded in the global optimal path record after the fitness calculation of the current candidate path is completed. If the fitness value of the current candidate path is less than the historical best fitness value, the global optimal path record is overwritten with the current candidate path and its corresponding B-spline control point sequence, and the current fitness value is updated to the new historical best fitness value. If the fitness value of the current candidate path is not less than the historical best fitness value, the global optimal path record remains unchanged. The global optimal path update process is executed uniformly after the fitness calculation of all candidate paths in each round of population iteration to ensure that the global optimal path always records the candidate path with the best overall quality that has appeared during the iteration process.

[0080] Furthermore, the trajectory generation module converts the globally optimal path into a joint space path for the robotic arm, and performs B-spline interpolation optimization again to generate the smooth joint trajectory that ultimately drives the robotic arm's motion. The specific method for this is as follows: The trajectory generation module's processing flow includes, in sequence: joint space transformation processing of the global optimal path, B-spline interpolation optimization processing of the joint space path, and time parameterization processing of smooth joint trajectories. These three processing steps are executed sequentially, ultimately generating a joint angle-time series that can directly drive the servo controllers of each joint of the robotic arm.

[0081] (1) Joint space transformation processing for the global optimal path: The joint space transformation process refers to the process of converting the sequence of Cartesian space path points on the globally optimal path into a sequence of joint angles of the robotic arm through inverse kinematics solution. Specifically, for each path point on the globally optimal path, with the position and orientation of the robotic arm's end effector in Cartesian space as the objective, the inverse kinematics solver of the robotic arm is invoked to calculate the joint angle vector that meets the pose requirements. When multiple solutions exist for inverse kinematics, the minimum change in joint angle is prioritized, and the joint angle vector with the smallest difference from the joint angles of adjacent path points is selected from the multiple solutions as the joint space representation of the current path point to ensure the continuity and monotonicity of the joint space path. For path points with no solution in inverse kinematics, resampling is performed by adjusting the interpolation density between adjacent path points until all path points can find a joint space solution that meets the requirements, thus obtaining the joint angle sequence corresponding to the globally optimal path.

[0082] (2) B-spline interpolation optimization of joint space path: The B-spline interpolation optimization process refers to the process of using the joint angle vectors in the joint angle sequence as control points, and performing B-spline curve fitting on each joint in the joint space to generate smooth interpolation curves for each joint angle with respect to path parameters. Specifically, the method involves: independently constructing cubic B-spline curves for each joint of the robotic arm; using the joint angles at each path point as interpolation nodes; calculating the node vectors of the B-spline using uniform parameterization or chord length parameterization methods; determining the control point sequence of the B-spline through least-squares fitting or interpolation; verifying the continuity of the first and second derivatives of the generated B-spline curves to ensure that the angular velocity and angular acceleration curves of each joint trajectory meet the continuity requirements; if the verification reveals a sudden change in angular velocity or angular acceleration in the B-spline curve of a certain joint within a certain interval, local optimization is performed by adjusting the control point positions or increasing the node density in that interval until the continuity requirements are met, thereby obtaining the smooth B-spline interpolation curves for each joint.

[0083] The equation of the B-spline curve is: θ_k(u) = Σ_{i=0}^{n} N_{i,p}(u) · q_{k,i},u ∈ [0, 1] Where θ_k(u) is the joint angle value of the k-th joint at parameter u, N_{i,p}(u) is the i-th p-th B-spline basis function, q_{k,i} is the angle value of the i-th control point of the B-spline curve of the k-th joint, n is the number of control points minus one, and p is the degree of the B-spline. In this module, cubic B-splines are used, i.e., p=3.

[0084] (3) Time parameterization of smooth joint trajectories: The time parameterization process refers to the process of converting the joint angle B-spline curve with path parameter u as the independent variable into a joint angle-time series with time t as the independent variable. Specifically, the method is as follows: First, based on the maximum allowable angular velocity constraints and maximum allowable angular acceleration constraints of each joint of the robotic arm, as well as the maximum linear velocity constraint of the end effector, the B-spline curve is time-optimally allocated, and the upper limit of the velocity constraint at each parameter point on the curve is calculated. Then, a time-optimal trajectory planning algorithm (such as trapezoidal velocity planning or S-curve velocity planning) is used to solve the mapping relationship from parameter u to time t. Next, based on the solved parameter-time mapping relationship, the target joint angle value of each joint is sampled at equal time intervals from the B-spline curve of each joint, with the servo control cycle as the sampling interval, to obtain the target joint angle value of each joint in each control cycle. Finally, the target joint angle value sequences of all joints are aligned according to the time step to generate a smooth joint trajectory array that drives the robotic arm motion. This array is then sent as the output of the trajectory generation module to the servo controller of each joint of the robotic arm for execution.

[0085] The generation of the smooth joint trajectory ensures the following quality indicators: First, the angular velocity and angular acceleration curves of each joint are continuously differentiable throughout the entire motion time, avoiding vibration of the robotic arm or joint impact caused by sudden velocity changes; second, the entire motion time is as short as possible while meeting the joint motion constraints, so as to ensure the motion efficiency of the robotic arm; third, there are no collision risk points in the trajectory, and the minimum distance between the robotic arm and obstacles in the workspace always meets the safety distance requirements throughout the entire motion process.

[0086] To verify the advantages of the proposed IESC-based robotic arm path planning method compared to current mainstream methods, simulation experiments were conducted using the DOA, FTTA, IRIME, ESC, and IRRT methods, along with the method of this invention, on the same map. Table 1 summarizes the planned path length and computation time of these methods in different environments. Figure 3 Simulation paths for the six methods are shown in both (a) and (b).

[0087] Table 1. Comparative Analysis of Experimental Results of DOA, FTTA, IRIME, ESC, IRRT, and IESC Methods

[0088] As shown in Table 1, compared to mainstream methods in the field of robotic arm path planning, the IESC method of this invention exhibits better performance and stability under various conditions. Simulation results are also available. Figure 6 , Figure 4 , Figure 5 The specific details are shown in the video. Figure 4 This is a comparison chart showing the adaptability of the DOA method, FTTA method, IRIME method, ESC method, IRRT method and the robotic arm path planning method of this invention in two environments. Figure 4 These are iterative curves of the DOA method, FTTA method, IRIME method, ESC method, IRRT method, and the robotic arm path planning method of this invention in two environments. Figure 6 This is a box plot comparing the DOA method, FTTA method, IRIME method, ESC method, IRRT method and the robotic arm path planning method of this invention in two environments.

[0089] Example 2 Secondly, such as Figure 7 As shown, this invention provides a robotic arm path planning method based on an improved escape optimization algorithm, applied to the aforementioned robotic arm path planning system based on an improved escape optimization algorithm, specifically including: S1 determines the iterative processing time of the robotic arm in the path planning process based on the path planning data, and determines the comparison and analysis processing time range of the image features of the working environment map based on the iterative processing time of different path iterative processing processes. S2 determines the acquisition and processing scheme of the identified image features by utilizing the similarity of image features within different comparison analysis time intervals; S3, based on the acquisition and processing scheme, performs the update processing of the recognition image features. According to the analysis results of the similarity of the recognition image features, the recognition image features are divided into different groups. Based on the recognition image features in the group and the deviation of the iteration processing time between the recognition image features, the target iteration number adjustment processing strategy in the group is determined. S4 adjusts the target iteration number in the group according to the adjustment processing strategy, and determines the target iteration number identification processing method according to the adjustment processing strategy of the target iteration number in different groups and the adjustment data in different iteration number intervals.

[0090] Furthermore, the path planning data of the robotic arm includes the number of path planning processes of the robotic arm and the iteration processing time of the path planning process.

[0091] Specifically, the method for determining the time interval for comparing and analyzing the image features of the working environment map is as follows: During iterative processing, the delay in the path iteration process is determined based on the iteration processing time in different path iteration processes. Based on the delay, the time interval for image feature comparison and analysis is determined, thus laying the foundation for optimizing and updating the target iteration number using the image feature comparison and analysis results.

[0092] The core objective of this embodiment is to determine the time interval for image feature comparison and analysis of the working environment map based on the path planning data of the robotic arm. Its core logic involves statistically analyzing the distribution characteristics of iteration processing time during different path iterations, identifying processes with excessively long iteration times, and extracting the intervals within which the iteration times of these processes are concentrated. This allows for determining the time interval range for image feature comparison and analysis. The overall logic follows the process of "iteration time distribution statistics → target interval proportion judgment → comparison and analysis processing time interval determination".

[0093] S11 determines the proportion of path iteration processes within different iteration processing time intervals based on the iteration processing time of different path iteration processes.

[0094] The path iteration process refers to each complete iteration optimization process performed by the robotic arm in a given working environment map when using an improved escape optimization algorithm for path planning. The iteration processing time refers to the time consumed by a single path iteration process from initializing the population to completing the target number of iterations and outputting the optimal path. The iteration processing time interval refers to dividing the numerical range of iteration time into multiple consecutive numerical intervals, each interval corresponding to a range of iteration times. The quantity ratio refers to the ratio of the number of path iteration processes within a certain iteration processing time interval to the total number of all path iteration processes.

[0095] Suppose the robotic arm performs multiple path iteration processes. Divide the iteration processing time of these processes into multiple iteration processing time intervals, count the number of path iteration processing processes falling into each interval, and calculate the proportion of path iteration processing processes in each interval.

[0096] This step provides a computational basis for determining the proportion of processes in each iteration processing time interval. Its significance lies in aggregating and statistically analyzing the scattered iteration processing time data according to intervals to form a quantitative description of the process distribution density in each interval. This provides data support for identifying the proportion of processes in intervals with longer iteration times and avoids making inaccurate judgments based solely on the iteration time value of a single process.

[0097] S12 determines the process proportion of the iterative processing time interval based on the proportion of the number of path iterative processing processes in different iterative processing time intervals.

[0098] The process percentage refers to the percentage of path iteration processes within a certain iteration processing time interval, expressed as a percentage. It is the product of the percentage of path iteration processes within that interval and the total number of intervals, reflecting the relative importance of that interval among all intervals.

[0099] Assuming the iteration processing time is divided into multiple intervals, and the percentage of path iteration processing processes in each interval has been statistically determined, then the percentage of processes in each interval is calculated sequentially to obtain the distribution of process percentages in each interval.

[0100] This step converts the quantity proportion within the interval into a unified process proportion index. Its significance lies in eliminating the influence of the number of interval divisions on the proportion results, making the process proportions between different intervals comparable, and providing a standardized basis for subsequent comparison analysis processing time intervals based on process proportions.

[0101] S13 determines the comparison and analysis processing time interval of the image features of the working environment map based on the process proportion in different iteration processing time intervals.

[0102] The image features refer to multi-dimensional feature information such as the type distribution of obstacles in the work environment map and the quantity composition of different types of obstacles. The comparison and analysis processing time interval refers to the iterative time interval for extracting and comparing the image features of the work environment map, so as to use the similarity of image features to determine whether the target iteration number is reasonable during the iterative processing corresponding to this interval.

[0103] Assuming that the process proportion of each iteration processing time interval has been obtained, based on the distribution of the process proportion, the intervals with longer iteration times and the intervals with higher process proportions are selected as the time intervals for image feature comparison analysis of the working environment map.

[0104] This step transforms the analysis results of the process proportion into specific comparison analysis processing time intervals. Its significance lies in establishing a mapping relationship between the distribution characteristics of iterative processing time and the requirements of image feature comparison analysis. This ensures that image feature comparison analysis is only performed on intervals with abnormal iteration time or concentrated process numbers, avoiding indiscriminate image feature comparison analysis on all intervals, thereby improving the overall comparison analysis efficiency.

[0105] It should be noted that the comparison and analysis processing time interval of the image features of the working environment map is determined by the proportion of the process in different iteration processing time intervals. Specifically, this includes taking the iteration processing time interval with an iteration time longer than the preset iteration time interval as the target interval.

[0106] The target interval refers to the iteration processing time interval where the iteration time exceeds the upper limit of the preset iteration time interval threshold. The path iteration processing within these intervals takes a long time, and there may be problems with low optimization efficiency due to unreasonable setting of the target iteration number. Therefore, these intervals are objects that need to be focused on and subjected to image feature comparison analysis.

[0107] Assuming the preset iteration duration interval threshold is a time value, the intervals above this threshold in each iteration processing duration interval are identified as target intervals. Then, the target intervals are the intervals with longer iteration durations.

[0108] The significance of using intervals with excessively long iteration times as target intervals is to prioritize image feature comparison analysis for intervals with abnormal iteration times, ensuring that analytical resources are concentrated on the intervals most likely to require adjustment of the target iteration count, thereby improving the relevance and efficiency of the analysis.

[0109] S131, determine whether the sum of the process proportions of different target intervals is greater than the preset proportion threshold. If yes, proceed to step S132. If no, determine that the comparison and analysis processing time interval of the image features of the working environment map is not required. At this time, the iterative processing speed is faster and there is no need to perform the identification processing of the optimal number of target iterations.

[0110] The preset percentage threshold refers to the threshold value for judging whether the sum of the process percentages in the target interval constitutes a significant impact. If the sum of the process percentages in the target interval is less than or equal to the threshold, it indicates that the number of processes with longer iteration times is small, and their proportion in the entire path planning process is limited. The overall speed of the iteration process is relatively fast, and the current setting value of the target iteration number can meet the efficiency requirements of path planning.

[0111] Assuming there are multiple target intervals, the process proportions of these target intervals are added together to obtain the sum of the process proportions of the target intervals. This sum is then compared with a preset proportion threshold. If the sum is not greater than the preset proportion threshold, it indicates that the proportion of processes with longer iteration times is not large, and the overall iteration processing speed is relatively fast. Therefore, there is no need to perform comparative analysis to determine the processing time interval.

[0112] This step determines whether further comparative analysis is needed by judging the sum of the proportions of the process in the target interval. Its significance is to avoid unnecessary image feature comparison analysis when the iterative processing speed is already fast, thereby saving computing resources. At the same time, it ensures that the comparison analysis process is only started when the number of processes with long iteration times reaches a certain scale.

[0113] S132 determines whether the average iterative processing time of different path iterative processing processes is greater than a preset time threshold. If not, the comparison analysis processing time interval of the image features of the working environment map is determined to be the comparison analysis processing time interval where the number of iterative processing processes is greater than a preset process number threshold. If yes, the comparison analysis processing time interval of the image features of the working environment map is determined to be the comparison analysis processing time interval where the number of iterative processing processes is greater than a preset process number threshold and the target interval.

[0114] The average iterative processing time refers to the arithmetic mean obtained by dividing the sum of the iterative processing times of all path iterative processing processes by the total number of path iterative processing processes, reflecting the concentration level of the overall iterative processing time. The preset time threshold is a threshold value for judging whether the overall iterative processing time is significantly too long. The preset process number threshold is a threshold value for judging whether the number of path iterative processing processes within a certain iterative processing time interval is sufficient.

[0115] Assuming the average iteration processing time of all path iteration processes has been calculated, if the average value is greater than the preset time threshold, it indicates that the overall iteration processing time is too long. In this case, the intervals with more processes than the preset process number threshold and the target interval are included in the comparison analysis processing time interval. If the average value is not greater than the preset time threshold, it indicates that the overall iteration time is acceptable, but there are some intervals with a concentrated number of processes. In this case, only the intervals with more processes than the preset process number threshold are included in the comparison analysis processing time interval.

[0116] This step, based on a comparison of the average overall iteration processing time with a preset time threshold, determines whether to include the target interval in the comparison analysis processing time interval. Its significance lies in distinguishing two different scenarios: one is a scenario where the overall iteration time is relatively long and the number of processes in individual intervals is concentrated, requiring comparison analysis of both the target interval and the concentrated process intervals; the other is a scenario where the overall iteration time is acceptable but the number of processes in individual intervals is concentrated, requiring comparison analysis only of the concentrated process intervals, thereby achieving adaptive differentiation and processing of abnormal patterns in different iteration times.

[0117] This embodiment, through steps S11 to S132, realizes the comparative analysis and processing time interval for determining image features of the working environment map based on robotic arm path planning data. Its core value lies in four aspects: First, by dividing the iterative processing time intervals and statistically analyzing their proportions, scattered iterative time data is aggregated into structured interval distribution information. Second, by identifying target intervals and judging the sum of process proportions, a graded evaluation mechanism for the degree of iterative time anomalies is established, avoiding unnecessary comparative analysis of processes with already fast iteration speeds. Third, by judging the average iterative processing time, a distinction is made between two abnormal modes: the overall iterative time level and the local interval process concentration. Fourth, through adaptive processing for different scenarios, it ensures that the determination of the comparative analysis and processing time intervals can cover processes with excessively long iteration times while avoiding waste of comparative analysis resources.

[0118] Furthermore, the method for determining the image feature acquisition and processing scheme is as follows: In this embodiment, based on the similarity of image features within different comparison analysis time intervals, the degree of deviation in the distribution and number of obstacles in the algorithm within different comparison analysis time intervals is determined. Based on the degree of deviation in the distribution and number of obstacles within different comparison analysis time intervals, the matching degree of the target iteration number in the current algorithm's path planning process is determined. The matching degree is then used to determine the image feature acquisition processing scheme, which also lays the foundation for further combining the image features to perform target iteration number recognition and update processing.

[0119] The core objective of this embodiment is to determine the image feature acquisition and processing scheme by utilizing the similarity of image features within different comparison analysis time intervals. Its core logic involves calculating the similarity of obstacle image features between path iteration processes within each comparison analysis time interval, identifying deviation intervals with abnormal similarity, and determining whether image feature acquisition processing is necessary and which image features to acquire based on the number and proportion of deviation intervals. The overall logic follows the process of "interval similarity calculation → deviation interval identification → determination of image feature acquisition scheme".

[0120] S21 determines the average value of the similarity of image features between path iteration processes within the comparison analysis time interval based on the similarity of image features within different comparison analysis time intervals.

[0121] The similarity of image features refers to the similarity between the working environment maps corresponding to two path iteration processes in terms of obstacle type distribution and the quantity of obstacles of different types. The higher the similarity, the closer the obstacle distribution of the working environment maps faced by the two processes is. The average value refers to the arithmetic mean of the similarity of image features between all path iteration processes within a certain comparison analysis time interval.

[0122] Suppose that a certain comparison analysis time interval contains multiple path iteration processing processes. Calculate the similarity of obstacle type distribution and obstacle quantity between any two processes in each process. Combine these to obtain the image feature similarity between each pair of processes. Then, take the average of all pairwise similarities to obtain the average image feature similarity between the path iteration processing processes in that interval.

[0123] This step provides a computational basis for determining the baseline similarity of each comparison analysis time interval. Its significance lies in quantifying the consistency of obstacle distribution in the working environment map within the interval by using the interval average similarity. The higher the similarity, the more similar the environment faced by the path planning within the interval is. The difference in iteration time is more likely to come from the deviation in the setting of the target iteration number rather than the difference in obstacle distribution.

[0124] S22 takes the average value of the similarity of image features between the path iteration processes within the comparison analysis time interval as the benchmark similarity of the comparison analysis time interval.

[0125] The benchmark similarity refers to the similarity used to measure the working environment between path iteration processes within a certain comparison analysis time interval. Figure 1 The benchmark reference value for consistency indicates that the higher the value, the more similar the working environment faced by each process within the interval; the lower the value, the greater the difference in the working environment faced by each process within the interval.

[0126] Assuming that the average value of the image feature similarity between the path iteration processes within each comparison analysis time interval has been calculated, then each average value is directly used as the benchmark similarity for the corresponding interval.

[0127] This step fixes the average similarity within the interval as a standardized benchmark similarity index. Its significance lies in providing a unified comparison benchmark for subsequent judgments on whether there are biased intervals, ensuring that the similarity levels between different intervals are comparable.

[0128] It should be noted that the image features between the path iteration processes are determined based on the type of obstacle and the similarity in the number of obstacles of different types.

[0129] The above steps include the following: S221 determines whether there is a comparison analysis time interval where the baseline similarity is less than the preset similarity threshold. If yes, proceed to step S222. If no, determine that the image feature acquisition processing scheme does not require image feature acquisition processing. At this time, the time deviation is mainly due to the deviation of the obstacle type.

[0130] The preset similarity threshold refers to the threshold value for judging whether the benchmark similarity is significantly low. If the benchmark similarity of all comparison analysis time intervals is not less than the threshold, it indicates that the distribution of obstacles in the working environment map in each interval has a high degree of consistency. At this time, the deviation in iteration time is mainly due to the natural deviation of obstacle type, rather than the unreasonable setting of the target iteration number.

[0131] Assuming that the baseline similarity of each comparison analysis time interval has been obtained, each baseline similarity is compared with the preset similarity threshold one by one. If all baseline similarities are not less than the preset similarity threshold, it indicates that the distribution of obstacles in each interval is consistent and the iteration time deviation is a normal fluctuation, and there is no need to perform image feature acquisition processing.

[0132] This step quickly determines whether image feature acquisition needs to be performed by comparing the baseline similarity with the preset similarity threshold. Its significance lies in eliminating scenarios where iteration time fluctuates due to natural differences in obstacle types, and avoiding ineffective image feature analysis and target iteration number adjustment in such scenarios.

[0133] S222 defines the comparison analysis time interval where the benchmark similarity is less than the preset similarity threshold as the deviation interval. It determines whether the number of deviation intervals is greater than the preset deviation interval number threshold. If so, it determines that the image feature acquisition processing scheme is to use the image features of the iterative processing process with an iteration processing time less than the preset iteration time value as the recognition image features. Based on the similarity with the recognition image features, it optimizes the recognition processing of the target iteration number. This achieves the recognition processing of the optimal target iteration number without causing the iteration time to be too long. If not, it proceeds to step S23.

[0134] The deviation interval refers to the comparison analysis time interval where the baseline similarity is lower than a preset similarity threshold. Within these intervals, the obstacle distribution varies significantly between path iteration processes. The preset deviation interval number threshold refers to the critical value for determining whether the number of deviation intervals is sufficient to support reliable deviation analysis. The preset iteration time value refers to the upper limit for selecting iterative processes with shorter iteration times as the source of image features for identification.

[0135] Assuming there are multiple deviation intervals, if the number of deviation intervals is greater than the preset threshold, it means that there are enough deviation intervals. The image features of the iterative processing process with a shorter iteration time can be selected as the recognition image features. The environmental features of these "fast iteration" processes can be used as a benchmark to identify and evaluate the iteration time deviation of other processes.

[0136] When there are enough deviation intervals, this step adopts a strategy based on the "fast iteration" process. Its significance lies in selecting the environmental image features corresponding to the process with the highest iteration efficiency as a reference standard, ensuring that the subsequent target iteration optimization and recognition processing will not lose its guiding value due to the excessive iteration time of the reference standard itself.

[0137] S23 determines the acquisition and processing scheme of the recognized image features based on the process proportion in different comparison analysis time intervals and the benchmark similarity.

[0138] The above steps include the following: S231 obtains the process proportion of different deviation intervals, and determines whether there is a deviation interval whose process proportion is greater than the preset process proportion threshold. If yes, proceed to step S232. If no, it is determined that the image feature acquisition processing scheme is not accurate enough at this time, and it is difficult to determine whether the iteration time is too long due to the unreasonable number of target iterations. Therefore, it is determined that there is no need to perform image feature acquisition processing.

[0139] The preset process proportion threshold refers to the threshold value for judging whether the process proportion of the deviation interval is significant enough. If there is no deviation interval with a process proportion exceeding the threshold, it means that although the deviation interval exists, the number of processes is small and not enough to support reliable recognition image feature analysis.

[0140] Assuming multiple deviation intervals have been identified, the process proportion corresponding to each deviation interval is obtained and compared with a preset process proportion threshold one by one. If the process proportion of all deviation intervals is not greater than the preset process proportion threshold, it indicates that the influence range of the deviation interval is limited. At this time, the recognition result is not accurate enough, and it is determined that there is no need to perform image feature acquisition processing.

[0141] This step verifies the scale of the process proportion in the deviation interval. Its significance lies in ensuring that the acquisition and processing of image features is only initiated when the process proportion in the deviation interval reaches a certain scale, thus avoiding unstable recognition results due to insufficient data.

[0142] S232 determines the basic weight value of the deviation interval based on the process proportion of different deviation intervals, and determines the corrected similarity weight value based on the basic weight value of different deviation intervals and the basic similarity. It then determines whether the corrected similarity weight value is greater than a preset weight threshold. If so, it determines that the image feature acquisition processing scheme is to use the image features of the iterative processing process with an iterative processing time less than a preset iterative processing time value as the image features of the recognition image. If not, it determines that the image feature acquisition processing scheme is to use the image features of the iterative processing process with an iterative processing time less than a second preset iterative processing time value (less than the preset iterative processing time value) as the image features of the recognition image.

[0143] The basic weight value refers to the weight value reflecting the relative importance of each deviation interval, determined based on the process proportion of each deviation interval. The larger the process proportion of a deviation interval, the larger its basic weight value. The corrected similarity weight value refers to the corrected weight index obtained by comprehensively calculating the basic weight value of each deviation interval and its benchmark similarity, used to measure the overall deviation degree and scale of influence of the deviation interval. The second iteration duration preset value refers to another upper limit of iteration duration, which is less than the iteration duration preset value. It is used to select an iterative processing process with a shorter iteration duration as the source of image features for recognition when the corrected similarity weight value is low.

[0144] Let N be the number of deviation intervals, P_i be the process proportion of the i-th deviation interval, and S_i be the benchmark similarity.

[0145] The basic weight value W_i for each deviation interval is determined based on the process proportion P_i; the larger the process proportion, the larger the basic weight value. The corrected similarity weight value = Σ(W_i × S_i), i = 1 to N.

[0146] Assuming multiple deviation intervals exist, a basic weight value is determined based on the process proportion of each deviation interval. The basic weight value of each deviation interval is multiplied by the baseline similarity of that interval, and the results are accumulated to obtain a corrected similarity weight value. If this value is greater than a preset weight threshold, the iterative processing process with the shorter iteration time is selected as the source of image features for recognition; if this value is not greater than the preset weight threshold, a more stringent upper limit on the iteration time is used to select the source of image features for recognition.

[0147] This step comprehensively evaluates the overall impact of the deviation interval by constructing a modified similarity weight value. Its significance lies in adaptively adjusting the selection strategy of the recognized image features according to the process proportion of the deviation interval and the degree of similarity deviation: when the similarity is high, a more lenient selection condition is adopted to include more reference information, and when the similarity is low, a more stringent selection condition is adopted to avoid affecting the iteration time.

[0148] Furthermore, the identified image features are divided into different groups, specifically including: Image features whose similarity to each other exceeds a preset similarity threshold are grouped into the same group.

[0149] Furthermore, the method for determining the adjustment strategy for the target iteration number in the group is as follows: In this embodiment, the risk level of updating the target iteration number using the recognized image features in the group is determined based on the number of recognized image features in the group and the deviation of the iteration processing time. Based on the risk level of updating the target iteration number using the recognized image features in the group, an adjustment processing strategy for the target iteration number is determined, thereby reducing the impact on the path optimization planning time while ensuring the efficiency of the target iteration number update recognition processing.

[0150] The core objective of this embodiment is to classify identified image features into different groups based on their recognition. Then, based on the number of identified image features within each group and the deviation in iteration processing time, an adjustment strategy for the target number of iterations within that group is determined. The core logic involves grouping features into different groups through similarity clustering. Within each group, risk indicators are evaluated in two dimensions: the number of features and the iteration time deviation rate. This comprehensive assessment determines the appropriate target iteration number adjustment strategy for that group. The overall logic follows the process of "similarity clustering to divide into groups → judging the number of features within each group → judging the iteration time deviation rate within each group → calculating the adjustment fit coefficient → determining the adjustment processing strategy."

[0151] The process of dividing the identified image features into different groups specifically includes: grouping identified image features whose similarity to each other is greater than a preset similarity threshold into the same group.

[0152] S31 determines the number of recognized image features in the group based on the recognized image features in the group.

[0153] The group refers to a set of recognized image features whose similarity to each other exceeds a preset similarity threshold. The working environment maps corresponding to the recognized image features within the same group exhibit high consistency in obstacle distribution. The preset image feature quantity threshold refers to the minimum scale requirement for determining whether the number of recognized image features in the group meets the minimum requirement for adjusting the target iteration count.

[0154] Suppose that the features of the recognized image are divided into multiple groups by similarity clustering, and the number of recognized image features contained in each group is counted to obtain the feature count of each group.

[0155] This step provides basic data for subsequent judgment on whether a group meets the conditions for adjustment. Its significance lies in quickly filtering out groups with insufficient data through the statistics of the number of features, and avoiding unreliable target iteration number adjustment processing in such groups.

[0156] In the above steps, if the number of recognized image features in the group is less than the preset threshold for the number of image features, then no matter how similar the new recognized image features are to the recognized image features in the group, there is no need to adjust the target iteration number.

[0157] If a group contains only one image feature and the preset threshold for the number of image features is 3, then the number of features in the group is insufficient to support reliable target iteration number adjustment processing, and no adjustment is made.

[0158] Furthermore, if the number of recognized image features in the group is not less than a preset image feature number threshold, then proceed to step S32.

[0159] S32 determines the deviation rate of iterative processing time between different recognized image features based on the deviation of iterative processing time between recognized image features in the group.

[0160] The deviation rate of the iterative processing time refers to the difference rate between the iterative processing times of the path iterative processing processes corresponding to any two recognized image features in the group. It is taken as the absolute value of the difference between the two iterative processing times divided by the larger iterative processing time, reflecting the degree of difference in iterative efficiency between the two processes.

[0161] Suppose a group contains multiple recognition image features. Calculate the deviation rate of the iteration processing time of the path iteration process corresponding to any two recognition image features, and obtain the set of all pairwise deviation rates within the group.

[0162] This step provides a quantitative basis for subsequent judgment of the consistency of iteration time within the group. Its significance lies in measuring the stability of the iteration efficiency of each process within the group through the deviation rate. The lower the deviation rate, the more stable the iteration time within the group, and the higher the reliability of the target iteration number adjustment based on the group.

[0163] Specifically, the above steps include the following: obtaining the deviation rate of the iteration processing time between different recognition image features in the group, determining whether the average deviation rate of the iteration processing time between different recognition image features in the group is greater than a preset deviation rate threshold; if so, no matter how similar the new recognition image feature is to the recognition image feature in the group, there is no need to adjust the target iteration number; otherwise, proceed to step S33.

[0164] The preset deviation rate threshold refers to the threshold value for judging whether the deviation rate of the iteration processing time within the group is too large. If the average deviation rate is greater than the threshold, it indicates that the iteration time of each process within the group fluctuates too much. Even if the obstacle distribution in the working environment map is similar, the difference in iteration time is still significant, indicating that the group is not suitable as the basis for adjusting the target iteration number.

[0165] Assuming that the set of all pairwise deviation rates within a certain group has been calculated, calculate the mean of these deviation rates. If the mean is greater than the preset deviation rate threshold, it indicates that the iteration time within the group is not stable enough, and no adjustment of the target number of iterations is performed. If the mean is not greater than the preset deviation rate threshold, it indicates that the iteration time within the group is relatively stable, and the adjustment of the adaptation coefficient in step S33 can be further performed.

[0166] This step verifies the consistency of iteration duration within a group by judging the mean deviation rate. Its significance is to ensure that the target number of iterations is adjusted only when the iteration duration within the group is stable, thus avoiding insufficient reliability of the adjustment strategy due to excessive fluctuations in iteration duration between processes within the group.

[0167] S33 determines the adjustment processing strategy for the target number of iterations in the group based on the number of recognized image features in the group and the deviation rate of the iteration processing time between different recognized image features in the group.

[0168] Furthermore, based on the number of recognized image features in the group and the deviation rate of iterative processing time between different recognized image features in the group, an adjustment strategy for the target number of iterations in the group is determined, specifically including: S331 determines the adjustment adaptation coefficient of the group based on the number of recognized image features in the group and the deviation rate of the iteration processing time between different recognized image features in the group. It then determines whether the adjustment adaptation coefficient of the group is greater than a preset adaptation coefficient threshold. If so, the adjustment processing strategy for the target iteration number in the group is determined to be a reliable adjustment strategy. As long as the similarity between the new recognized image feature and the recognized image features in the group is greater than the preset similarity value, the target iteration number adjustment processing is performed in the path planning processing process corresponding to the new recognized image feature. Otherwise, proceed to step S332.

[0169] The adjustment fit coefficient refers to a fit index that comprehensively reflects the quantity and stability of the identified image features in the group and the iteration time, and is used to determine whether the group is suitable as a reliable basis for adjusting the target iteration number. The reliable adjustment strategy refers to a strategy that can directly perform target iteration number adjustment processing after meeting the basic similarity conditions, indicating that the group data is sufficient and stable, and the adjustment processing has high reliability.

[0170] Let M be the number of recognized image features in the group, and D_avg be the mean deviation rate of the iteration processing time between different recognized image features in the group. The adjustment adaptation coefficient = M / N ÷ (1 + D_avg), where N is the baseline quantity used for normalization.

[0171] Suppose a group contains multiple recognition image features. The average value of the iterative processing time deviation rate within the group has been calculated, and the adjustment adaptation coefficient is obtained. If the coefficient is greater than the preset adaptation coefficient threshold, a reliable adjustment strategy is adopted.

[0172] This step integrates information from two dimensions—feature quantity and bias rate—into a unified fitness index by constructing an adjustment fitness coefficient. Its significance lies in taking into account both the sufficiency of the group's samples and the stability of the iteration duration, ensuring that the most lenient and reliable adjustment strategy is adopted only when both aspects perform well.

[0173] S332 determines whether, during the path planning process corresponding to the recognized image features in the group, there is a path planning process where the average deviation rate of the iteration processing time between different recognized image features in the group is greater than a preset deviation rate threshold. If not, the adjustment strategy for the target iteration number in the group is determined to be a reliable adjustment strategy. If yes, the adjustment strategy for the target iteration number in the group is determined to be a general adjustment strategy. That is, as long as the similarity between the new recognized image feature and the recognized image features in the group is greater than a preset similarity value, and the similarity with the recognized image features that account for the proportion of the target quantity in the group is greater than a preset similarity threshold (greater than the preset similarity value), the target iteration number adjustment process is performed during the path planning process corresponding to the new recognized image feature.

[0174] The general adjustment strategy refers to adding an extra constraint to the similarity condition based on determining whether there are individual processes where the average deviation rate exceeds the standard. The target iteration count can only be adjusted after the similarity to the recognized image features representing a certain percentage of the target's quantity in the group reaches a required level. The target quantity percentage refers to a certain proportion of the recognized image features in the group.

[0175] Assuming the adjustment fit coefficient of a certain group is not greater than a preset fit coefficient threshold, further determine whether there are individual processes in the group whose average deviation rate exceeds the preset deviation rate threshold. If no such process exists, it indicates that the group has good overall stability, and a reliable adjustment strategy is adopted; if such a process exists, it indicates that there are individual processes in the group with excessive deviation, requiring additional similarity constraints, and a general adjustment strategy is adopted.

[0176] This step further refines the judgment when the fit coefficient is insufficient. Its significance lies in distinguishing whether there are individual "abnormal processes" in the group that cause the overall fit to decrease. If there are no abnormal processes but the fit coefficient is still insufficient, the sample size may be too small but the stability is good, and a reliable adjustment strategy can still be given. If there are abnormal processes, more stringent similarity conditions need to be added to ensure the reliability of the adjustment process.

[0177] The image features are identified as the image features of four working environment maps corresponding to P1, P2, P7, and P12. The similarity between the four identified image features is calculated as follows: P1-P2 = 0.88, P1-P7 = 0.85, P1-P12 = 0.90, P2-P7 = 0.86, P2-P12 = 0.82, and P7-P12 = 0.78. The preset similarity threshold is 0.80. The similarity between P1 and P2 is 0.88 > 0.80, the similarity between P1 and P7 is 0.85 > 0.80, the similarity between P1 and P12 is 0.90 > 0.80, the similarity between P2 and P7 is 0.86 > 0.80, the similarity between P2 and P12 is 0.82 > 0.80, and the similarity between P7 and P12 is 0.78, which is not greater than 0.80. Therefore, the similarity between P1, P2, and P12 is greater than 0.80, while the similarity between P7 and P12 is not greater than 0.80. After cluster analysis, P1, P2, and P12 are assigned to group G1, and P7 is assigned to group G2.

[0178] Group G1 contains 3 recognized image features. The preset threshold for the number of image features is 2. Since the number of recognized image features in group G1 is not less than 2, proceed to step S32. The iteration processing time for each recognized image feature in G1 is as follows: P1 is 80 seconds, P2 is 70 seconds, and P12 is 90 seconds. The deviation rate between P1 and P2 = |80-70|÷80 = 10÷80 = 0.125, the deviation rate between P1 and P12 = |80-90|÷90 = 10÷90 = 0.111, and the deviation rate between P2 and P12 = |70-90|÷90 = 20÷90 = 0.222. The average deviation rate = (0.125+0.111+0.222)÷3 = 0.458÷3 = 0.153. The preset deviation rate threshold is 0.30, and the average deviation rate of 0.153 is not greater than 0.30. Proceed to step S33.

[0179] In group G1, the number of image features identified is M=3, and the average deviation rate of iteration processing time is D_avg = 0.153.

[0180] The adaptation coefficient is adjusted to 3 / 10 ÷ (1 + 0.153) = 0.26. The preset adaptation coefficient threshold is 0.250. Therefore, the adjustment strategy for the target iteration number in group G1 is determined to be a reliable adjustment strategy. That is, as long as the similarity between the new recognition image feature and the recognition image features P1, P2, and P12 in group G1 is greater than the preset similarity value (set to 0.80), the target iteration number will be adjusted during the path planning process corresponding to the new recognition image feature.

[0181] Group G2 contains only one image feature P7. The number of features 1 is less than the preset threshold of the number of image features 2. Therefore, no matter how similar the new image feature is to the image feature P7 in group G2, there is no need to adjust the number of target iterations.

[0182] This completes the determination of the group division and target iteration number adjustment strategy: group G1 adopts a reliable adjustment strategy, while group G2 is not adjusted.

[0183] Furthermore, the adjustment of the target iteration count in the group according to the adjustment processing strategy specifically includes: In different groups, the target iteration number is adjusted once by using a preset number of recognized image features in all iteration intervals. Specifically, the target iteration number is adjusted in different iteration intervals in different groups in a loop. That is, the benchmark iteration number in the iteration interval is used as the target iteration number, where the benchmark iteration number can be the average of the endpoints of the iteration interval.

[0184] The iteration number interval refers to dividing the target iteration number range into multiple consecutive sub-intervals, each sub-interval corresponding to a range of target iteration numbers. The baseline iteration number refers to a reference value used to represent the target iteration number within each iteration number interval, taken as the arithmetic mean of the upper and lower endpoints of the interval. The preset quantity refers to the number of recognized image features used in each adjustment process.

[0185] Specifically, the method for determining the target iteration number is as follows: In this embodiment, the efficiency of the current group in identifying the target iteration number is determined by using the adjustment processing strategy for the target iteration number in different groups and the adjustment data in different iteration number intervals. Based on the efficiency of the current group in identifying the target iteration number using the adjustment processing strategy, the method for identifying the target iteration number is determined, that is, which iteration number intervals are eliminated, thereby further improving the efficiency and reliability of the target iteration number identification process.

[0186] S41, using the adjustment processing strategy for the number of target iterations in different groups, determine the adjustment time weight value for the number of target iterations in different groups.

[0187] The adjustment timeliness weight value refers to the weight value that reflects the timeliness of the adjustment process of a group, determined according to the type of adjustment processing strategy adopted by the group. The adjustment timeliness weight value of a reliable adjustment strategy is greater than that of a general adjustment strategy, indicating that the group corresponding to the reliable adjustment strategy has better timeliness performance in adjustment processing.

[0188] Assuming there are multiple groups, each group is assigned a corresponding adjustment timeliness weight value based on its adjustment handling strategy type: groups using reliable adjustment strategies are assigned higher adjustment timeliness weight values, while groups using general adjustment strategies are assigned lower adjustment timeliness weight values.

[0189] This step provides basic weight data for subsequent assessment of adjustment timeliness and elimination judgment within the iteration range. Its significance lies in distinguishing the differences in adjustment timeliness caused by different adjustment strategies among different groups, so that subsequent assessments can take into account the impact of strategy reliability on adjustment efficiency.

[0190] The above steps include the following: Based on the adjustment time weight value of the target iteration number in different groups, determine whether the average value of the adjustment time weight value of the target iteration number in different groups is less than a preset weight threshold. If so, as long as the average iteration processing time in any group is greater than the average iteration processing time of the group before the iteration number adjustment processing in the iteration number interval, and the deviation rate of the average iteration processing time of the group before the iteration number adjustment processing in the iteration number interval is greater than a preset deviation rate threshold, then the iteration number interval is excluded, that is, the target iteration number adjustment processing is no longer performed in the iteration number interval. If not, proceed to step S42.

[0191] The preset weight threshold refers to the threshold value for judging whether the average level of the adjustment time weight value of each group is high enough. If the average value is low, it means that the reliability of the overall adjustment strategy is low. Therefore, a more lenient standard is adopted for the elimination conditions in the iteration number interval - the interval can be excluded as long as any one group meets the elimination conditions.

[0192] Assuming the adjustment timeliness weight values ​​for each group have been determined, calculate their average value. If the average value is less than the preset weight threshold, it indicates that the overall adjustment timeliness is insufficient. In this case, as long as the average post-adjustment iteration processing time of any group within a certain iteration interval is greater than that before adjustment and the deviation rate exceeds the threshold, that iteration interval can be excluded.

[0193] This step sets different levels of strictness for elimination criteria based on the overall adjustment timeliness weight value. Its significance lies in using a more lenient elimination standard when the overall strategy reliability is low, quickly eliminating the poorly performing iteration range, and avoiding the continued consumption of computing resources under an unstable strategy.

[0194] S42 determines the number of adjustments in different groups within different iteration intervals based on the adjustment data within those intervals.

[0195] The number of adjustments refers to the number of times a group actually performs the target number of iterations adjustment processing based on the newly matched recognition image features within a certain interval of iterations, reflecting the degree of active adjustment of the group within that interval.

[0196] Assuming there are multiple iteration intervals and multiple groups, we count the number of times each group actually underwent adjustment processing within each interval, and obtain the distribution matrix of the number of adjustments.

[0197] This step provides a data foundation for subsequent identification of groups with insufficient adjustment activity and deviations in the number of iterations. Its significance lies in discovering groups with abnormal adjustment numbers within a specific range through statistical analysis of the number of adjustments, thus providing a basis for evaluating the overall applicability of the iteration number range.

[0198] The above steps include the following: determining the groups whose adjustment counts within different groups in different iteration count intervals are less than a preset adjustment count threshold, identifying these groups as iteration count deviation groups, determining whether there exists an iteration count interval with an iteration count deviation group, and if so, proceeding to step S43; otherwise, if the average iteration processing time in any group is greater than the average iteration processing time of the group before iteration count adjustment processing within the iteration count interval, and the deviation rate of the group from the average iteration processing time of the group before iteration count adjustment processing within the iteration count interval is greater than a preset deviation rate threshold, then the iteration count interval is excluded, i.e., the target iteration count adjustment processing is no longer performed within the iteration count interval.

[0199] The iteration number deviation group refers to the group whose number of adjustments is lower than the preset adjustment number threshold within a certain iteration number interval. The adjustment activity of these groups in this interval is insufficient, indicating that some intervals may not have been fully verified, and therefore further identification and processing are required.

[0200] Assuming the number of adjustments for each group within each interval has been counted, for each interval, groups with an adjustment count less than a preset adjustment count threshold are selected as the iteration count deviation groups for that interval. If there is no iteration count interval with an iteration count deviation group, it means that all intervals have been reliably verified, that is, the number of adjustments for all groups within all intervals has met the requirements. In this case, the same elimination condition corresponding to "average value less than threshold" in S41 is used.

[0201] This step further refines the evaluation of the iteration number range by identifying the deviation groups of the iteration number. Its significance lies in using the group with insufficient adjustment number as a condition for judging whether different intervals have been sufficiently verified and identified. When there is no deviation group, a lenient elimination standard is adopted, and when there is a deviation group, it enters the more refined S43 evaluation.

[0202] S43 determines the identification and processing method of the target iteration number by using the adjustment time weight value of the target iteration number in different groups and the adjustment number in different groups within different iteration number intervals.

[0203] The above steps include the following: S431 determines whether the time for adjusting the target number of iterations is greater than a preset time threshold. If yes, proceed to step S432. If no, determine that the method for identifying the target number of iterations is that only when the average of the iteration processing time of all groups is greater than the average of the iteration processing time of the group before the iteration number adjustment processing in the iteration number interval, and the deviation rate of the group from the average of the iteration processing time of the group before the iteration number adjustment processing in the iteration number interval is greater than a preset deviation rate threshold, then the iteration number interval is excluded, that is, the target number of iterations is no longer adjusted in the iteration number interval.

[0204] The duration of the adjustment process for the target number of iterations refers to the total time elapsed from the start of the group loop adjustment process to the current moment, reflecting the duration of the adjustment iteration process. The preset duration threshold is a threshold value used to determine whether the adjustment process has been running for a sufficient amount of time.

[0205] Assuming the target number of iterations has been adjusted, calculate the total time from the start of the adjustment to the present. If this time is not greater than the preset time threshold, it indicates that the adjustment process was too short and data accumulation may be insufficient. In this case, the strictest elimination condition is adopted—the iteration number range can only be excluded when all groups meet the elimination condition.

[0206] This step sets different levels of strictness for elimination conditions based on the cumulative duration of the adjustment process. Its significance lies in using the strictest conditions when the adjustment process time is short and the data accumulation is insufficient to avoid prematurely eliminating potentially effective iteration ranges and ensuring that the identification and processing of the target iteration number has sufficient data support.

[0207] S432 determines the adjustment adaptation coefficient based on the average of the adjustment time weight values ​​of the target iteration number in different groups and the proportion of iteration number intervals of groups without iteration number deviation. It then determines whether the adjustment adaptation coefficient is greater than a preset adaptation coefficient threshold. If so, the iteration number interval is excluded if the average iteration processing time in multiple groups is greater than the average iteration processing time of the group before iteration number adjustment processing within the iteration number interval, and the deviation rate from the average iteration processing time of the group before iteration number adjustment processing within the iteration number interval is greater than a preset deviation rate threshold. That is, the target iteration number adjustment processing is no longer performed within the iteration number interval. If not, the target iteration number identification processing method is determined to be that the iteration number interval is excluded only if the average iteration processing time of all groups is greater than the average iteration processing time of the group before iteration number adjustment processing within the iteration number interval, and the deviation rate from the average iteration processing time of the group before iteration number adjustment processing within the iteration number interval is greater than a preset deviation rate threshold. That is, the target iteration number adjustment processing is no longer performed within the iteration number interval.

[0208] The adjustment fit coefficient refers to an adjustment index that comprehensively reflects the timeliness and applicability of the adjustment range, and is used to determine the strictness of the elimination conditions for the iteration number range. The proportion of iteration number ranges without iteration number deviation groups refers to the proportion of all iteration number ranges without iteration number deviation groups to the total number of ranges.

[0209] Let W_avg be the mean of the adjustment time weight values ​​for the target iteration number in different groups, and R be the proportion of iteration number intervals in groups without iteration number deviation. The adjustment fit coefficient = (1 - W_avg + R) / 2.

[0210] It should be noted that the smaller the average value of the adjustment time weight of the target iteration number in the different groups, the higher the proportion of the number of iteration number intervals of the groups without iteration number deviation, the larger the adjustment adaptation coefficient, which reflects the sufficiency of verification and the efficiency of identification processing. That is, the lower the efficiency of identification processing, the higher the sufficiency of verification, and the more lenient the exclusion conditions.

[0211] Assuming the mean of the adjustment timeliness weight values ​​for each group has been calculated and the proportion of intervals without iteration number deviation has been obtained, the adjustment fit coefficient is obtained. If the coefficient is greater than the preset fit coefficient threshold, it indicates that the overall adjustment conditions are good, and a more lenient elimination condition is adopted (multiple groups can be excluded if they meet the condition); if the coefficient is not greater than the preset fit coefficient threshold, it indicates that the overall adjustment conditions are average, and a more stringent elimination condition is adopted (all groups must meet the condition to be excluded).

[0212] This step determines the final elimination conditions by comprehensively evaluating the overall adjustment status through adjusting the adaptation coefficient. Its significance lies in integrating the group's strategy reliability index and interval applicability index into a comprehensive judgment standard, realizing adaptive adjustment of the interval elimination strategy for the number of iterations, and ensuring that interval elimination can be carried out with an appropriate degree of strictness under different overall conditions.

[0213] Specifically, the optimal value of the target number of iterations is the base number of the interval in which the average of the average iteration processing time in all groups is less than the average iteration processing time before the iteration number adjustment process, and the deviation rate between the average iteration processing time of the group and the average iteration processing time before the iteration number adjustment process within the interval is the largest.

[0214] This embodiment determines the method for identifying the target iteration number through steps S41 to S432. Its core value lies in three aspects: First, by introducing the adjustment timeliness weight value, it distinguishes the impact of different strategies' reliability on adjustment evaluation; second, by analyzing the distribution of adjustment number, it identifies groups and intervals with abnormal adjustment activity, providing data basis for interval elimination; and third, by adaptively setting multi-level elimination conditions, it achieves elimination processing of iteration number intervals with an appropriate degree of strictness under different adjustment states, ensuring that the identification of the optimal target iteration number is neither too aggressive nor too conservative.

[0215] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0216] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0217] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A robotic arm path planning system based on an improved escape optimization algorithm, characterized in that, Specifically, it includes: The model building module is responsible for acquiring the working environment map and performing 3D modeling to obtain the environment map model and boundary constraints. The population generation module is responsible for generating the original path population based on the boundary constraints using random initialization, and performing adaptive Gaussian perturbation on a portion of the selected original path population to generate a new path population. The original path population and the new path population are merged and optimized to form the initial path set. The iterative optimization module is responsible for updating the paths in the initial path set into calm groups, panic groups, and conformist groups based on the fitness of the paths. Specifically, the calm and panic groups are updated based on a spiral search strategy combined with a dynamic panic index, while the conformist group is updated based on the attraction to the center of the calm group and the global optimum. After the number of iterations reaches the target number of iterations, a dynamic mutation mechanism is introduced to improve the ability of candidate paths to escape local optima. A multi-objective fitness function is constructed, which includes path length, collision penalty, path height, and path smoothness. Collision detection and fitness calculation are performed on the smooth trajectories generated by B-spline interpolation of candidate paths, and the global optimum path is updated based on the calculation results. The trajectory generation module is responsible for converting the global optimal path into a joint space path of the robotic arm, and then performing B-spline interpolation optimization to generate a smooth joint trajectory that ultimately drives the robotic arm to move.

2. The robotic arm path planning system based on the improved escape optimization algorithm as described in claim 1, characterized in that, The boundary constraints are constructed using the space-occupying markers of static obstacles in the working environment map.

3. The robotic arm path planning system based on the improved escape optimization algorithm as described in claim 1, characterized in that, The original path population and the new path population are merged and optimized to form the initial path set, specifically including: Based on the preset population size and search space, and in conjunction with the boundary constraints, the original path population is generated using random initialization. From the original path population, a preset number of original valid paths are randomly selected as base paths, and a new path population is generated by performing multiple Gaussian perturbations on each selected base path. The original path population and the new path population are merged, their fitness is calculated and sorted from low to high, and the top-ranked populations are selected as the final initial paths.

4. The robotic arm path planning system based on the improved escape optimization algorithm as described in claim 1, characterized in that, The paths in the initial path set are divided into three groups—calm group, panic group, and conformity group—based on their fitness for updating. Specifically, this includes: The paths in the initial path set are sorted in ascending order of fitness, and the paths are divided into calm group, panic group and conformity group according to a preset ratio; The paths of the calm group and the panic group are updated based on the spiral search strategy of the whale optimization algorithm, and the panic index is introduced as a random perturbation term. The path of the conformist group is updated based on its own inertia, the center of the calm group, and the attraction of the global optimal solution, and a boundary adaptive exploration mechanism is introduced based on the boundary gravity.

5. The robotic arm path planning system based on the improved escape optimization algorithm as described in claim 1, characterized in that, The dynamic mutation mechanism includes uniform spherical mutation and jump mutation.

6. The robotic arm path planning system based on the improved escape optimization algorithm as described in claim 5, characterized in that, The radius of the uniform and jump variations of the sphere is dynamically adjusted based on the panic index and the diagonal length of the map size.

7. A robotic arm path planning method based on an improved escape optimization algorithm, applied to the robotic arm path planning system based on an improved escape optimization algorithm as described in any one of claims 1-6, characterized in that, Specifically, it includes: Based on the path planning data of the robotic arm, the iterative processing time of the robotic arm in the path planning process is determined. Based on the iterative processing time of different path iterative processing processes, the comparison and analysis processing time range of image features of the working environment map is determined. The similarity of image features within different comparison analysis time intervals is used to determine the acquisition and processing scheme of the identified image features; Based on the acquisition and processing scheme, the image features of the recognition are updated. According to the analysis results of the similarity of the image features, the image features are divided into different groups. Based on the deviation of the iterative processing time of the image features in the group and the recognition image features, the target iteration number adjustment processing strategy in the group is determined. The target iteration number in the group is adjusted according to the adjustment processing strategy, and the target iteration number identification processing method is determined based on the adjustment processing strategy for the target iteration number in different groups and the adjustment data in different iteration number ranges.

8. The robotic arm path planning method based on the improved escape optimization algorithm as described in claim 7, characterized in that, The path planning data of the robotic arm includes the number of path planning processes of the robotic arm and the iteration processing time of the path planning process.

9. The robotic arm path planning method based on the improved escape optimization algorithm as described in claim 7, characterized in that, The method for determining the time interval for image feature comparison and analysis processing of the working environment map is as follows: Based on the iteration processing time in different path iteration processing processes, determine the proportion of the number of path iteration processing processes in different iteration processing time intervals; The process proportion of the iterative processing time interval is determined based on the proportion of the number of path iterative processing processes in different iterative processing time intervals. The comparison and analysis processing time interval of the image features of the working environment map is determined by the proportion of the process in different iteration processing time intervals.

10. The robotic arm path planning method based on the improved escape optimization algorithm as described in claim 7, characterized in that, The method for determining the target iteration number is as follows: S41 determines the adjustment timeliness weight value of the target iteration number in different groups based on the adjustment processing strategy for the target iteration number in different groups; S42 determines the number of adjustments in different groups within different iteration intervals based on the adjustment data within those intervals; S43 determines the identification and processing method of the target iteration number by using the adjustment time weight value of the target iteration number in different groups and the adjustment number in different groups within different iteration number intervals.

Images