Path planning method, device, equipment and medium based on particle filter optimization

Abstract

The present application discloses a path planning method, device, equipment and medium based on particle filter optimization, and relates to the field of path planning technology. The method comprises: initializing a group of particles according to the state vector of the target robot at the current moment, each particle represents the state vector sample of the target robot at the next moment; the state vector comprises position and speed; based on the initialized group of particles, using an improved particle filter optimization algorithm to determine the state vector at the next moment; the improvement of the improved particle filter optimization algorithm comprises that after resampling, if the iteration stop condition is not met, a conversion kernel method is introduced to move particles in the particle set to obtain an updated particle set, a K-means clustering algorithm is used to cluster particles in the updated particle set to obtain a cluster-optimized particle set, and whether the iteration stop condition is met is determined according to the cluster-optimized particle set. The present application improves the reliability of path search.

Description

Technical Field

[0001] The present application relates to the technical field of path planning, and in particular to a path planning method, device, equipment and medium based on particle filter optimization. Background Art

[0002] Positioning is one of the most critical performances in autonomous mobile robots and vehicle navigation. The key to achieving autonomous mobile robots and vehicle navigation is to solve the positioning problem. To ensure the safety of mobile robots or self-driving cars, a rigorous and reliable positioning algorithm is essential. In a complex real environment, sensor noise or environmental noise imposes limitations on the quality of sensor readings, so the positioning algorithm cannot correctly deduce the actual state of the environment. Therefore, a particle filter is introduced as a state estimator to improve accuracy. However, the conventional particle filter algorithm will have insufficient particle diversity in some areas, which will affect the reliability of state estimation. Summary of the invention

[0003] The purpose of this application is to provide a path planning method, device, equipment and medium based on particle filter optimization, which can improve the reliability of path search.

[0004] To achieve the above-mentioned purpose, the present application provides the following solutions: In the first aspect, the present application provides a path planning method based on particle filter optimization, including: according to the state vector of the target robot at the current moment, initialize a group of particles, each particle represents a state vector sample of the target robot at the next moment; the state vector includes position and speed; based on the initialized group of particles, use an improved particle filter optimization algorithm to determine the state vector at the next moment; the improvement of the improved particle filter optimization algorithm includes that after resampling, if the iteration stop condition is not met, a conversion kernel method is introduced to move the particles in the particle set to obtain an updated particle set, and the K-means clustering algorithm is used to cluster the particles in the updated particle set to obtain a cluster-optimized particle set, and whether the iteration stop condition is met is judged according to the cluster-optimized particle set; the cluster-optimized particle set includes each particle group and the center of mass of each particle group.

[0005] In a second aspect, the present application provides a path planning device based on particle filter optimization, and the path planning device based on particle filter optimization includes: an initialization module, which is used to initialize a group of particles according to the state vector of the target robot at the current moment, and each particle represents a state vector sample of the target robot at the next moment; the state vector includes position and speed; a state vector prediction module, which is used to determine the state vector at the next moment based on the initialized group of particles, using an improved particle filter optimization algorithm; the improvement of the improved particle filter optimization algorithm includes that after resampling, if the iteration stop condition is not met, a conversion kernel method is introduced to move the particles in the particle set to obtain an updated particle set, and the K-means clustering algorithm is used to cluster the particles in the updated particle set to obtain a cluster-optimized particle set, and whether the iteration stop condition is met is judged according to the cluster-optimized particle set; the cluster-optimized particle set includes each particle group and the center of mass of each particle group.

[0006] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any one of the above-mentioned path planning methods based on particle filter optimization.

[0007] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned path planning methods based on particle filter optimization.

[0008] According to the specific embodiments provided by the present application, the present application discloses the following technical effects: the present application provides a path planning method, device, equipment and medium based on particle filter optimization. In the improved particle filter optimization (PFO) algorithm, after resampling, if the iteration stop condition is not met, a conversion kernel method is introduced to move the particles in the particle set to obtain an updated particle set, and a K-means clustering algorithm is used to cluster the particles in the updated particle set to obtain a cluster-optimized particle set, thereby improving the uniformity of particle distribution in the particle set, judging whether the iteration stop condition is met based on the cluster-optimized particle set, optimizing the judgment criterion of the stop condition, solving the problem of failure to converge to the true state due to uneven particle distribution, and improving the reliability of path search. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0010] Figure 1 A schematic flowchart of a path planning method based on particle filter optimization is provided in one embodiment of the present application.

[0011] Figure 2 A schematic diagram of the flow chart of an improved particle filter optimization algorithm provided in one embodiment of the present application.

[0012] Figure 3 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0014] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0015] Particle filters can estimate the internal state of a dynamic system from a series of noisy or incomplete observations.

[0016] System state estimation is the basis for the application of autonomous mobile robots. Dynamic path planning is a research hotspot in the field of robotics. In a dynamic environment, the target path from the starting point to the end point is modified online and in real time according to the evaluation index. Existing path planning methods include graph search algorithms and intelligent optimization algorithms. The real-time performance of graph search algorithms decreases with the increase of the environment scale. Intelligent optimization algorithms are robust and have considerable adaptability, but they will converge prematurely and have poor local search. In dynamic and complex environments, they all have their own limitations. As a state estimation method suitable for strongly nonlinear and non-Gaussian systems, particle filters are non-parametric heuristic algorithms that use recursive sampling to model the probability space, a set of samples extracted from the proposed distribution to approximate the system state distribution, and are mainly used for time prediction and estimation. When the signal is non-Gaussian and nonlinear, it is usually impossible to use standard estimators such as minimum mean square error, which usually leads to poor solutions. Therefore, the particle filter is considered to be introduced as a state estimator to improve the accuracy. Conventional particle filter algorithms will have insufficient particle diversity in some areas, which will affect the accuracy and reliability of the estimation. Insufficient number of particles, uneven distribution of particle weights, and inappropriate resampling strategies will cause ordinary particle filter algorithms to fail to converge to the true state, or to estimate the state inaccurately, resulting in poor robustness. Conventional particle filter algorithms show slow convergence to the optimal value and low search efficiency.

[0017] This application provides a path planning method based on particle filter optimization, such as Figure 1 As shown, the present application provides a path planning method based on a particle filter, including steps 101 to 102.

[0018] Step 101: Initialize a group of particles according to the state vector of the target robot at the current moment, each particle represents a state vector sample of the target robot at the next moment; the state vector includes position and velocity.

[0019] Each particle is a point in the robot's current state space.

[0020] Step 102: Based on a group of initialized particles, an improved particle filter optimization algorithm is used to determine the state vector at the next moment; the improvement of the improved particle filter optimization algorithm includes introducing a conversion kernel method to move particles in the particle set to obtain an updated particle set if the iteration stop condition is not met after resampling, clustering the particles in the updated particle set using a K-means clustering algorithm to obtain a cluster-optimized particle set, and judging whether the iteration stop condition is met based on the cluster-optimized particle set; the cluster-optimized particle set includes each particle group and the centroid of each particle group.

[0021] In the improved particle filter optimization algorithm of the present application, after resampling, if the iteration stop condition is not met, a conversion kernel method is introduced to move particles in the particle set to obtain an updated particle set, and a K-means clustering algorithm is used to cluster the particles in the updated particle set to obtain a cluster-optimized particle set, thereby improving the uniformity of particle distribution in the particle set, judging whether the iteration stop condition is met based on the cluster-optimized particle set, optimizing the judgment criterion of the stop condition, solving the problem of failure to converge to the true state due to uneven particle distribution, and improving the reliability of path search.

[0022] In this application, the probabilistic method of particle filters is transplanted to the field of optimization by converting the optimization problem into a filtering problem. In this case, the weight of the particles is determined according to the objective function value of the candidate solution in each iteration. Similar to the particle filter (PF), the possibility of incorporating uncertainty into the algorithm makes particle filter optimization a sampling-based technique with great potential in robust optimization problems. In order to improve the efficiency of robust optimization search and obtain local and global optimal solutions, it is worth knowing and better understanding the shape of the objective function.

[0023] A clustering algorithm is an unsupervised learning method used to divide samples in a data set into several different groups, so that the similarity between samples in each group is high, while the similarity between samples in different groups is low. The goal of a clustering algorithm is to discover the intrinsic structure in the data and divide the data into subsets with similar characteristics. The K-Means Clustering (KM) algorithm used in this application divides the samples in the data set into K clusters, and iteratively optimizes the position of the center point of each cluster to minimize the distance from each sample to the center point of its cluster.

[0024] The traditional PF architecture consists of three steps: the first step is the prediction step, which is responsible for moving the sample elements called particles forward one time step according to the state transition function of the state space model; the second step is the correction step, in which the particles obtain weights according to their consistency with the measured noise observations, and the probability is calculated using the measurement function; the third step is the re-adoption step, in order to ensure that the particle set is sufficiently effective.

[0025] The improved particle filter optimization algorithm of this application can be summarized as a maximization task as shown in formula (1).

[0026] (1)

[0027] in, is the evaluation function of the path to be planned for the autonomous mobile robot, is the target robot state vector, is the exponential coefficient matrix, is the constraint condition within the time range. Formula (1) is the path evaluation function. k is the number of iterations, Specifically, they are the energy cost on the path and the cost constraints of collision obstacles.

[0028] The optimization problem of formula (1) is transformed into formula (2) and formula (3), which are iterative filtering described by the state space model. Formula (2) corresponds to the prediction step in PF, and formula (3) corresponds to the correction step in PF. Filtering refers to estimating and optimizing the state in an iterative manner, thereby gradually improving the credibility of the true state.

[0029] (2)

[0030] in, is the candidate solution for the kth iteration, is the solution obtained by resampling at the k-1th iteration. The particles are classified according to their relative The objective function value is weighted.

[0031] (3)

[0032] in, is the noise observation value in the kth iteration, is the observation noise at the kth iteration, For The evaluation value for evaluation, is the optimal particle at the kth iteration, that is, the optimal solution. When Assign the value of .

[0033] In the initial stage of the search task, a feasible solution space is constructed based on the current position of the robot and the reference point, and a set of random samples is initially generated. The reference point refers to the position information of the key nodes on the path and the boundary information of the obstacles. The initial sample set is generated using orthogonalization, and then the weight of the sample is updated according to the path evaluation function and formula (4). Formula (4) is a continuous representation of the weighted particle distribution in the current iteration, which determines the calculation method of the weight. Indicates the number of particles. represents the distribution of the observation noise and determines how particles are distinguished during reweighting in the correction step. , then The value is set to Therefore, as the iteration proceeds, the performance interval of particles with non-zero weights becomes smaller and smaller. The nature of the function has not changed, and it is constrained by the lowest and highest objective function values ​​of the current iteration, so it actually explores smaller and smaller regions of the search space and concentrates around the optimum.

[0034] (4)

[0035] in, is the continuous representation of the distribution of weighted particle i in the kth iteration, that is, the target probability density function, is the weight of particle i in the kth iteration, is the number of particles, represents the distribution of observation noise, is particle i in the kth iteration, is the noise observation value in the kth iteration, It is the basis for weighting candidate solutions and selecting better solutions. Express The evaluation value for evaluation, is the weight of particle i in the k-1th iteration.

[0036] The PFO algorithm process is as follows: step a to step e.

[0037] Step a: In the initial stage of the search task, a feasible solution space is constructed based on the current position of the target robot and the reference point, a set of random samples is initially generated, and an initial sample set, i.e., a set of particles, is generated using orthogonalization.

[0038] Step b, use formula (2) to convert Transformed into a state space model description.

[0039] Step c: Get the observation noise using formula (5) and calculate the observation value using formula (3) (Occurs during the iteration , then The value is set to ), and then update the sample weight according to the path evaluation function and formula (4); is particle i in the kth iteration, for The weight of .

[0040] Step d, get the local optimal solution , update the local path; is the optimal solution, specifically refers to the state vector at the next moment.

[0041] Step e: resample at the next moment position and reference point to generate a new sample space, and repeat step b until the stop criterion is met. When the stop criterion is met, the end point is reached.

[0042] The distribution of the observation noise is shown in formula (5).

[0043] (5)

[0044] (6)

[0045] in, is the observation noise at the kth iteration, is the intermediate parameter, is the weight ratio of the best particle to the worst particle in the current particle set, is the lowest objective function value in the current particle set at the kth iteration, is the highest objective function value in the current particle set at the kth iteration, and the objective function is the evaluation function of the path to be planned for the target robot.

[0046] The best particle in the current particle set is the particle with the largest weight value, and the worst particle in the current particle set is the particle with the smallest weight value.

[0047] The iteration stop condition is shown in formula (7).

[0048] (7)

[0049] in, is the objective function of particle i in the kth iteration, is the objective function of particle i in the k-1th iteration, represents the expected value, is the threshold value.

[0050] The K-means clustering algorithm is used to cluster the particles in the updated particle set, and the clustering optimization objective is shown in formula (8).

[0051] (8)

[0052] Among them, K is the number of cluster centers, is the center of mass of the jth particle group, is particle i in the kth iteration, express and The squared Euclidean distance, is the jth particle group.

[0053] The process of K-means clustering algorithm includes: Step 1: Select the number of clusters K; Step 2: Define the initial centroid; Step 3: Consider the squared Euclidean distance metric and assign the data points to the nearest centroid; Step 4: Recalculate the centroid as the average of the data points in the cluster; Step 5: If the centroid is the same as the previous iteration, stop: otherwise return to step 3 to continue iteration.

[0054] The improvement of the improved particle filter optimization algorithm also includes introducing random variables in the prediction step of the state vector. The prediction formula after the introduction of the random variables is shown in formula (9).

[0055] (9)

[0056] in, is the candidate solution for the kth iteration, is the solution obtained by resampling at the k-1th iteration, is the random variable of the kth iteration, the candidate solution and the solution are both state vectors, and k is an integer greater than 1.

[0057] In the standard case, the new particle ( ) is obtained from the old particle ( ), so after several iterations, the particles will be concentrated on several points in the search space (even points in the initial population). In order to solve the problem of sample poverty, the improved PFO algorithm proposed in this application uses both replacement resampling and the conversion kernel method. The conversion kernel method is shown in formula (9). By introducing a random variable The selection probability of each particle is proportional to its normalized weight. In order to diversify the particles, random variables are introduced. ,random variable Extracted from the transition kernel symmetric to zero, the transition kernel is used to describe the probability distribution of particle state transition, and thus used to update the particle state according to the dynamic model of the system. After the introduction of the transition kernel, the selection probability of each particle is proportional to its normalized weight, and the modified state of the particle is shown in formula (9). At the beginning of the optimization, the random variable A higher value is set to ensure that the entire search space is searched, and the random variable Set to a low value to prevent particles from leaving areas with high target values.

[0058] because The function only gives values ​​greater than zero in the positive domain, so only those that perform better than The particles with positive weight will have positive weight, while the other particles will have zero weight. In the resampling step, only particles with non-zero weights will be propagated to the next iteration. The common method to distinguish particles with non-zero weights is to use uniform distribution. This will cause the weights of particles with better performance to be consistent with the weights of general particles, and it is impossible to accurately distinguish particles with different performances. In our improved algorithm, formula (5) and formula (6) are used to improve this situation. , Represents the lowest and highest objective function values ​​in the current particle set at the Kth iteration. Adjustable, Indicates the weight ratio of the best particle to the worst particle in the current particle set. Good particle: Its observation value is better than , with positive weights; particles with a weight of 0 are bad particles, and different weights are assigned to good particles and general particles using formulas (5) and (6).

[0059] In the conventional PFO algorithm termination criterion, if no better solution is found in the last ten iterations, the algorithm will terminate, which may not contain information about the entire particle set. Therefore, this application redefines the criterion when the optimization algorithm ends, considering defining the termination condition based on the expected value of the particle, as shown in formula (7).

[0060] After obtaining the final set of particles, the question arises how to evaluate the results. The global optimum can be defined as the particle with the best performance in the last iteration or in the iteration. However, it is more satisfactory to determine the average value of the particles based on the particle position only when it is determined that all particles are clustered around the global optimal point. This application introduces the K-means clustering algorithm, and the clustering optimization objective is shown in formula (8). In the clustering process, the centroid of each group The particle set will move and assign the data point to the nearest centroid point. By clustering the particle set, information about the relationship between the decision variables and the target value can be obtained: along which variables can more groups be identified, which variables are concentrated around a point, and how the variables converge one by one can also be analyzed. The improved PFO algorithm flow chart is as follows Figure 2 shown.

[0061] like Figure 2 As shown, the present application combines the K-means clustering algorithm to change the discrimination criteria at the end of the optimization algorithm, realizes local optimization robustness evaluation, finds the global optimum and the local optimum in a single run, and thus improves the robustness of the algorithm. In addition, the present application also solves the problem of sample poverty by converting the kernel method, improves robustness and search efficiency, and achieves better search efficiency. The present application reduces the computing requirements without losing particle performance, improves search efficiency, and obtains the global optimum and the local optimum. The efficiency of the path planning search process of the autonomous mobile robot in a dynamic environment is improved, and the estimation accuracy is improved.

[0062] Based on the same inventive concept, the embodiment of the present application also provides a particle filter-optimized path planning device for implementing the particle filter-optimized path planning method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more particle filter-optimized path planning device embodiments provided below can refer to the limitations of the particle filter-optimized path planning method above, and will not be repeated here.

[0063] In an exemplary embodiment, the present application provides a path planning device based on particle filter optimization, which includes an initialization module and a state vector prediction module.

[0064] The initialization module is used to initialize a group of particles according to the state vector of the target robot at the current moment, each particle represents a state vector sample of the target robot at the next moment; the state vector includes position and speed.

[0065] The state vector prediction module is used to determine the state vector at the next moment based on an initialized group of particles by using an improved particle filter optimization algorithm; the improvement of the improved particle filter optimization algorithm includes introducing a conversion kernel method to move particles in the particle set to obtain an updated particle set if the iteration stop condition is not met after resampling, clustering the particles in the updated particle set by using a K-means clustering algorithm to obtain a cluster-optimized particle set, and judging whether the iteration stop condition is met according to the cluster-optimized particle set; the cluster-optimized particle set includes each particle group and the centroid of each particle group.

[0066] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store path planning data based on particle filter optimization. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a path planning method based on particle filter optimization is implemented.

[0067] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0068] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0069] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0070] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0071] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0072] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a data processing logic of a programmable logic device, etc., but is not limited thereto.

[0073] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0074] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A path planning method based on particle filter optimization, characterized in that: The particle filter-based path planning method comprises: Initialize a group of particles according to the state vector of the target robot at the current moment, each particle represents a state vector sample of the target robot at the next moment; the state vector includes position and velocity; Based on an initialized group of particles, an improved particle filter optimization algorithm is used to determine the state vector at the next moment; the improvement of the improved particle filter optimization algorithm includes introducing a conversion kernel method to move particles in a particle set to obtain an updated particle set if the iteration stop condition is not met after resampling, clustering the particles in the updated particle set using a K-means clustering algorithm to obtain a cluster-optimized particle set, and judging whether the iteration stop condition is met according to the cluster-optimized particle set; the cluster-optimized particle set includes each particle group and the centroid of each particle group: The formula for updating the weight of particles in the correction step of the improved particle filter optimization algorithm is: ; in, is the weight of particle i in the kth iteration, is the number of particles, represents the distribution of observation noise, is particle i in the kth iteration, is the noise observation value in the kth iteration, Express The evaluation value for evaluation, is the weight of particle i in the k-1th iteration; The distribution of the observation noise is expressed as: , ; in, is the observation noise at the kth iteration, is the intermediate parameter, is the weight ratio of the best particle to the worst particle in the current particle set, is the lowest objective function value in the current particle set at the kth iteration, is the highest objective function value in the current particle set at the kth iteration, and the objective function is the evaluation function of the path to be planned of the target robot; The calculation formula for the noise observation value in the kth iteration is: ; in, is the observation noise at the kth iteration, For The evaluation value for evaluation, is the optimal particle at the kth iteration. When Assign the value of .

2. The path planning method based on particle filter optimization according to claim 1, characterized in that: The improvement of the improved particle filter optimization algorithm also includes introducing random variables in the prediction step of the state vector, wherein the prediction formula is expressed as: ; in, is the candidate solution for the kth iteration, is the solution obtained by resampling at the k-1th iteration, is the random variable of the kth iteration, the candidate solution and the solution are both state vectors, and k is an integer greater than 1.

3. The path planning method based on particle filter optimization according to claim 1, characterized in that: The iteration stop condition is expressed as: ; in, is the objective function of particle i in the kth iteration, is the objective function of particle i in the k-1th iteration, represents the expected value, is the threshold value.

4. The path planning method based on particle filter optimization according to claim 1, characterized in that: The clustering optimization objective of clustering the particles in the updated particle set using the K-means clustering algorithm is expressed as: ; Among them, K is the number of cluster centers, is the center of mass of the jth particle group, is particle i in the kth iteration, express and The squared Euclidean distance, is the jth particle group.

5. A path planning device based on particle filter optimization for implementing the path planning method based on particle filter optimization according to any one of claims 1 to 4, characterized in that: The path planning device based on particle filter optimization comprises: An initialization module, used to initialize a group of particles according to the state vector of the target robot at the current moment, each particle representing a state vector sample of the target robot at the next moment; the state vector includes position and velocity; The state vector prediction module is used to determine the state vector at the next moment based on an initialized group of particles by using an improved particle filter optimization algorithm; the improvement of the improved particle filter optimization algorithm includes introducing a conversion kernel method to move particles in the particle set to obtain an updated particle set if the iteration stop condition is not met after resampling, clustering the particles in the updated particle set by using a K-means clustering algorithm to obtain a cluster-optimized particle set, and judging whether the iteration stop condition is met according to the cluster-optimized particle set; the cluster-optimized particle set includes each particle group and the centroid of each particle group.

6. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the path planning method based on particle filter optimization according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the path planning method based on particle filter optimization described in any one of claims 1 to 4 is implemented.

Images