A radioactive source search method and system

Abstract

The application discloses a kind of radioactive source searching method and system, belong to robot searching technical field, the method includes: the grid map corresponding to the search environment is constructed;Based on the grid map, the radiation data of radioactive source is observed using sensor, and observation data is obtained;Using improved particle filtering algorithm, observation data is processed to estimate the position of radioactive source;According to the estimated radioactive source position and grid map, the travel path of robot is navigated;According to particle distribution, it is judged whether to terminate radioactive source search.The technical scheme of the present application can improve the success rate of radioactive source search, increase the effectiveness and robustness of the task of performing radioactive source search.

Description

Technical Field

[0001] This invention relates to the field of robot search technology, and in particular to a method and system for searching for radioactive sources. Background Technology

[0002] With the continuous advancement of technology, robotics has played a vital role in various fields, especially in hazardous environments. The search and location of radioactive sources is a crucial task in modern society, involving multiple areas such as nuclear facility safety inspections, environmental monitoring, and emergency response. Radioactive sources emit powerful penetrating rays and are widely used in medicine, agriculture, security, and other fields. However, radioactive sources also possess a double-edged sword characteristic: leakage or illegal dissemination of radioactive sources can pose a significant threat to human health and environmental safety. Therefore, how to quickly, accurately, and effectively detect and locate radioactive sources has become an urgent problem to be solved. To address this challenge, the application of robotics, especially mobile robot platforms, in radioactive source search has received increasing attention.

[0003] The complexity of radioactive source search tasks lies in several aspects. First, radioactive sources typically emit weak radiation signals, requiring detectors with sufficient sensitivity. Second, the location of radioactive sources may be unknown, or even hidden in complex environments such as confined spaces or underground passages. Furthermore, radioactive sources may exist in dynamic environments, with potential interference from other sources, making traditional manual search methods inefficient, inaccurate, and posing significant safety risks. Therefore, developing an automated and intelligent radioactive source search system is crucial to solving this problem.

[0004] Robotic mobile platforms, as intelligent devices capable of autonomous navigation and task execution in complex environments, possess immense potential in radioactive source search. By integrating sensors, positioning systems, and intelligent algorithms, robots can automatically plan paths, avoid obstacles, and search for and locate radioactive sources without human intervention. Existing research and technologies have made some progress in certain applications, but still face numerous challenges and limitations.

[0005] Despite the enormous potential of robotics in radioactive source search missions, it still faces many challenges in practical applications:

[0006] Environmental complexity: Radioactive sources may exist in extremely complex environments, especially in confined or irregular spaces. Robots need to be able to autonomously cope with various complex factors in the environment, such as obstacles, narrow passages, and different terrains.

[0007] Weak signal and high interference: The radiation signals emitted by radioactive sources are usually weak, and there may be overlap between signals from multiple sources. In addition, there may be other types of interference sources in the environment (such as electromagnetic radiation, temperature changes, etc.), all of which can affect the sensitivity of the detector and the accuracy of positioning.

[0008] Localization and Navigation: Accurate localization and navigation are crucial for autonomous robots. The location of radiation sources may be uncertain, and the robot may be affected by dynamic changes in the environment over time (e.g., moving obstacles, electromagnetic interference, etc.). Therefore, the robot needs efficient real-time path planning and adaptive adjustment capabilities.

[0009] Multi-sensor fusion: To improve search efficiency and accuracy, existing technologies typically employ combinations of multiple sensors, such as radiation detectors, lidar, and infrared sensors. However, data from different types of sensors often require fusion and processing to obtain more accurate results. Achieving efficient fusion of multi-sensor data and making informed decisions based on this fusion is a major challenge in technological development.

[0010] In conclusion, robotic mobile platforms have broad application prospects in radioactive source search, but still face some technical challenges. Traditional radioactive source localization methods suffer from low search efficiency and are greatly affected by obstacles. Summary of the Invention

[0011] This invention provides a method and system for searching radioactive sources, in order to solve the technical problems of low efficiency and great influence of obstacles in traditional radioactive source localization.

[0012] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0013] On one hand, the present invention provides a method for searching for radioactive sources, comprising:

[0014] Construct a grid map corresponding to the search environment;

[0015] Based on the grid map, radiation data from the radioactive source is observed using sensors to obtain observation data;

[0016] An improved particle filter algorithm was used to process the observation data to estimate the location of the radiation source;

[0017] The robot's path is navigated based on the estimated location of the radiation source and the grid map.

[0018] The decision to terminate the search for the radioactive source is based on the particle distribution.

[0019] Furthermore, the improved particle filtering algorithm filters historical observation data through the farthest point sampling algorithm, merges the filtered historical observation data with the current observation data, constructs a joint likelihood probability density using the merged observation data, and updates the particle weights based on the constructed joint likelihood probability density.

[0020] Furthermore, the process of using the improved particle filter algorithm to process the observation data to estimate the location of the radiation source includes:

[0021] Define a weighted set of particles of size N, {p i ,w i ∨i=0,1,2,…,N}, complete particle initialization; where p i This represents the state of the i-th particle. x i y i Indicates the particle position. Indicates the radiation intensity of a particle; w i This represents the weight of the i-th particle, and the initial value of the weight of each particle is 1 / N, and a uniform distribution is used as the initial distribution of the particles.

[0022] After obtaining the k-th observation data z k Then, the historical observation data is filtered using the farthest point sampling algorithm to select m valid historical observation data. Based on the selected historical observation data and z... k Construct the joint likelihood probability density F w The formula is:

[0023]

[0024] Where, q i P(z) represents the probability weight; i |x k P(z) represents the likelihood probability density of historical observations; k |x k ) represents the likelihood probability density of the current observation;

[0025] The particle weights are updated based on the constructed joint likelihood probability density.

[0026] During particle filtering, while keeping the total number of particles constant, particles with larger weights are replicated more often, while particles with smaller weights are replicated less often. Here, particles with larger weights refer to particles with weights greater than a preset threshold, and particles with smaller weights refer to particles with weights no greater than the preset threshold.

[0027] During the convergence process of particle filtering, the resampled particles are adaptively updated. The adaptive update method is to add Gaussian perturbation to the particles and distribute them according to Gaussian distribution with the original state as the expected value.

[0028] In the final stage of particle filtering, the average state of all particles is calculated as the predicted value for this round of filtering.

[0029] Furthermore, the step of determining whether to terminate the search for the radioactive source based on the particle distribution includes:

[0030] The convergence of the particle distribution is calculated. The mean-shift algorithm is used to perform cluster analysis on the particle distribution. If the distribution radius of the particles is smaller than the preset radius value, the search is terminated; otherwise, the search continues.

[0031] On the other hand, the present invention also provides a radioactive source searching system, comprising:

[0032] The data acquisition module is used for:

[0033] Construct a grid map corresponding to the search environment;

[0034] Based on the grid map, radiation data from the radioactive source is observed using sensors to obtain observation data;

[0035] The data processing module is used for:

[0036] An improved particle filter algorithm was used to process the observation data to estimate the location of the radiation source;

[0037] The robot's path is navigated based on the estimated location of the radiation source and the grid map.

[0038] The decision to terminate the search for the radioactive source is based on the particle distribution.

[0039] Furthermore, the improved particle filtering algorithm filters historical observation data through the farthest point sampling algorithm, merges the filtered historical observation data with the current observation data, constructs a joint likelihood probability density using the merged observation data, and updates the particle weights based on the constructed joint likelihood probability density.

[0040] Furthermore, the process of using the improved particle filter algorithm to process the observation data to estimate the location of the radiation source includes:

[0041] Define a weighted set of particles of size N, {p i ,w i ∨i=0,1,2,…,N}, complete particle initialization; where p i This represents the state of the i-th particle. xi y i Indicates the particle position. Indicates the radiation intensity of a particle; w i This represents the weight of the i-th particle, and the initial value of the weight of each particle is 1 / N, and a uniform distribution is used as the initial distribution of the particles.

[0042] After obtaining the k-th observation data z k Then, the historical observation data is filtered using the farthest point sampling algorithm to select m valid historical observation data. Based on the selected historical observation data and z... k Construct the joint likelihood probability density F w The formula is:

[0043]

[0044] Where, q i P(z) represents the probability weight; i |x k P(z) represents the likelihood probability density of historical observations; k |x k ) represents the likelihood probability density of the current observation;

[0045] The particle weights are updated based on the constructed joint likelihood probability density.

[0046] During particle filtering, while keeping the total number of particles constant, particles with larger weights are replicated more often, while particles with smaller weights are replicated less often. Here, particles with larger weights refer to particles with weights greater than a preset threshold, and particles with smaller weights refer to particles with weights no greater than the preset threshold.

[0047] During the convergence process of particle filtering, the resampled particles are adaptively updated. The adaptive update method is to add Gaussian perturbation to the particles and distribute them according to Gaussian distribution with the original state as the expected value.

[0048] In the final stage of particle filtering, the average state of all particles is calculated as the predicted value for this round of filtering.

[0049] Furthermore, the step of determining whether to terminate the search for the radioactive source based on the particle distribution includes:

[0050] The convergence of the particle distribution is calculated. The mean-shift algorithm is used to perform cluster analysis on the particle distribution. If the distribution radius of the particles is smaller than the preset radius value, the search is terminated; otherwise, the search continues.

[0051] The beneficial effects of the technical solution provided by this invention include at least the following:

[0052] This invention innovatively introduces the Farthest Point Sampling (FPS) algorithm into the traditional particle filter algorithm. FPS is used to manage historical observation data, thereby enabling the fusion of different observation data. This fusion method reduces noise interference affecting the traditional particle filter algorithm, improving its convergence speed and robustness. Consequently, it increases the success rate of radioactive source search, enhancing the effectiveness and robustness of radioactive source search tasks. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a flowchart of the radioactive source searching method provided in the embodiments of the present invention;

[0055] Figure 2 This is a schematic diagram of the real-world modeling and simulation provided in the embodiments of the present invention using Gazebo;

[0056] Figure 3 This is a schematic diagram of the Gazebo robot building platform and sensor configuration provided in an embodiment of the present invention;

[0057] Figure 4 This is a schematic diagram of the move_base navigation framework provided in an embodiment of the present invention;

[0058] Figure 5 This is a schematic diagram of the overall particle filter framework provided in an embodiment of the present invention;

[0059] Figure 6 This is a schematic diagram of the roulette wheel betting method provided in an embodiment of the present invention;

[0060] Figure 7 This is a schematic diagram of the particle filter search process provided in an embodiment of the present invention;

[0061] Figure 8 This is a schematic diagram illustrating particle clustering convergence and termination of the search provided in an embodiment of the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0063] First, it should be noted that in the embodiments of the present invention, the words "exemplarily," "for example," etc., are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the term "exemplarily" is intended to present the concept in a specific manner. Furthermore, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either one or the other.

[0064] This embodiment provides a method for searching for radioactive sources. This method can be implemented by an electronic device, which can be a terminal or a server. The execution flow of this method is as follows: Figure 1 As shown, it includes the following steps:

[0065] S1, Construct a grid map corresponding to the search environment;

[0066] It should be noted that in SLAM systems, a grid map is constructed to represent the occupancy status in the environment. The grid map is represented by a two-dimensional matrix M, where each element corresponds to a grid cell in the environment. Assuming the environment is a two-dimensional plane, and the environment is constructed as a two-dimensional grid plane with a size of m x n, that is, the grid map has a size of m x n, then the mathematical representation of the grid map is a two-dimensional array: M = [m...]. ij Among them, m ij This represents the occupancy probability of grid cell (i,j). Each grid cell m in the grid map... ij Typically, a probability value is used to represent whether a grid cell is occupied by an obstacle. The probabilistic model in this invention is based on Bayesian filtering. In this model, the probability P(m) of each cell in the grid map... ij The occupancy probability () indicates the likelihood that the cell is occupied by an obstacle. This probability can be updated through sensor observations and robot movement.

[0067] Occupancy probability P(m) ij The following conditions must be met:

[0068]

[0069] Wherein, P(m) ij The value represents the probability that the grid cell is occupied by an obstacle. There is usually an intermediate value to represent the case where the position is uncertain.

[0070] S2, Based on the grid map, the radiation data of the radioactive source is observed using a sensor to obtain observation data;

[0071] It should be noted that the platform used in this invention is a mobile robot equipped with a single radiation sensor. Based on this, this invention utilizes a sparse dataset to locate and determine the intensity of the radiation source when gradient direction information is unavailable. The Geiger-Muller tube used in this invention has high measurement efficiency, small size and mass, and is easy to install on a small unmanned reconnaissance platform. Since the measured information is the cumulative radiation dose of omnidirectional gamma rays, the estimation model proposed in this invention is needed to estimate the location and intensity of the radiation source. The observation probability model of this sensor adopts a Poisson distribution to characterize the probability of a counting event with an expected count rate of λ and an actual count rate of c per unit time. The Poisson distribution probability is shown below:

[0072]

[0073] Where λ represents the expected number of times the statistical event occurs per unit time, and c represents the actual number of times the statistical event occurs.

[0074] S3 uses an improved particle filter algorithm to process the observation data in order to estimate the location of the radiation source;

[0075] It should be noted that this invention innovatively introduces the farthest point sampling algorithm (FPS) into the particle filter (PF) algorithm. The FPS algorithm is used to manage historical observation data and fuse historical observations with current observation information to improve the classic particle filter (PF) algorithm. Through fusion, a new likelihood probability model can be constructed based on the observation model in particle filtering to update the particle weights. This can reduce the interference of noise on the traditional particle filter algorithm and improve the convergence speed and robustness of the particle filter algorithm.

[0076] The core idea of ​​the FPS algorithm is to make the observed data as discrete and uniform as possible by sampling from the farthest point. First, Euclidean distance is used to measure the distance between observation points. Then, the centroid of the point set is calculated, and the point farthest from the centroid is selected and denoted as P0. Next, from the remaining points, the point farthest from P0 is selected and denoted as P1. For each remaining point, the distance to both P0 and P1 is calculated, and the shortest distance is selected as the overall distance from that point to P0 and P1. After calculating these distances, the point with the largest distance is selected and denoted as P2. This process is repeated until the required number of points are selected.

[0077] Particle filtering mainly includes: particle initialization, weight update, particle resampling, and adaptive update.

[0078] Specifically, in this embodiment, S3 includes the following steps:

[0079] S31, First, in the particle initialization part, define a weighted particle set {p} of size N. i ,w i∨i=0,1,2,…,N};where, the state of the particle And the initial weight of each particle is 1 / N, x i y i Indicates the particle position. This represents the radiation intensity of the particles; the initial distribution of the particles is chosen based on prior experience. This invention uses a uniform distribution as the initial distribution of the particles, meaning that the probability of each map grid cell containing a radiation source is equivalent, and the formula satisfies:

[0080]

[0081] S32, Next is the update of the particle weights, after obtaining the k-th observation z. k Then, based on the previous FPS sliding window algorithm, m valid historical observations are selected and compared with the current z. k A joint likelihood probability density is constructed, and this density is used as the weight for particle updates. The formula for the joint likelihood probability density is:

[0082]

[0083] Where, q i P(z) represents the probability weight; i |x k P(z) represents the likelihood probability density of historical observations; k |x k ) represents the likelihood probability density of the current observation;

[0084] According to Bayesian theory, the state transition function of the particle weights is equivalent to the joint likelihood probability, thus completing the update of the particle weights.

[0085] S33. In particle filtering, particle degradation often occurs, meaning that during iterative updates, only a very small number of particles have significant weights, while the weights of the majority of particles are very small or even close to zero. This results in a very large variance in particle weights, leading to a significant waste of computational resources on updating useless and redundant particles. To avoid particle degradation, particle resampling is necessary. While keeping the total number of particles constant, particles with larger weights are replicated more frequently, while particles with smaller weights are replicated less often.

[0086] S34. During the convergence process of particle filtering, the search range continuously decreases, making it susceptible to noise and prone to getting trapped in local optima. Therefore, adaptive updates are needed for the resampled particles to improve their search capability. The adaptive update method involves adding Gaussian perturbations to the particles and distributing them according to a Gaussian distribution with the original state as the expected value.

[0087] In step S35, the average state of all particles is calculated as the predicted value for this round of filtering in the final stage of particle filtering. This allows for the estimation of the location and intensity of the radiation source based on the particle filtering algorithm, the calculation of the robot's motion direction vector, and then, combined with the grid map state, navigation of the robot's path.

[0088] S4. Navigate the robot's path based on the estimated location of the radiation source and the grid map.

[0089] It should be noted that after estimating the location and intensity of the radiation source, the robot can be controlled to navigate to the target point calculated by S3, thus completing the navigation of the robot's travel path.

[0090] S5, determine whether to terminate the search for the radioactive source based on the particle distribution;

[0091] It should be noted that this invention determines whether to terminate the search for the radioactive source based on the particle distribution. If the particle distribution convergence reaches a certain threshold, the search stops. Specifically, the convergence of the particle distribution is calculated, and the distribution of the particle set is clustered using the mean-shift algorithm. If the distribution radius of the particles is less than 3m, the search terminates. Otherwise, it returns to S2 and continues the search.

[0092] The effectiveness of the method of the present invention will be verified by simulation below.

[0093] like Figure 2 As shown, this invention uses the ROS middleware and the Gazebo simulation platform together. Gazebo can create a simulation scene for testing robots, imitate the real world by adding an object library, and add 2D environment design drawings through the Building Editor to build 3D houses based on the design drawings.

[0094] like Figure 3 As shown, this invention utilizes the Gazebo sensor model library to build a four-wheeled robot platform, including a Mecanum omnidirectional chassis, and is equipped with commonly used robot sensors such as a camera, depth camera, laser, and IMU. Furthermore, a radiation sensor is created, its specific parameters are added, and Gaussian noise is added as the sensor noise model.

[0095] like Figure 4As shown, this invention leverages the MoveBase open-source algorithm framework to enable robot simulation components. It uses the Gmaping algorithm for laser SLAM mapping, the AMCL algorithm for indoor localization, the A* algorithm for global path planning to give the robot navigation capabilities, and the Teb algorithm for local path planning to give the robot obstacle avoidance capabilities. The robot navigation process can be visualized in real-time on the RViz platform.

[0096] This embodiment assumes that there is a radioactive point source in a certain area whose location and radioactivity are unknown, and considers a two-dimensional planar area while ignoring the vertical height of the radioactive source.

[0097] Vector g k =[x k ,y k ,z k ] indicates that in (x k ,y k One observation point at location z k This represents the radiation count at the observation point. X = [x] s ,y s ,I s ] represents the radioactive source parameter vector, (x s ,y s ) represents the two-dimensional coordinate position of the radioactive source, I s This refers to the activity information of the radioactive source.

[0098] During nuclear decay, the count values ​​in a radiation counter follow a Poisson distribution. If the radiation count rate at a certain location is μ, then the probability that a radiation sensor detects a count value z within time τ follows a Poisson distribution. Where λ = μτ is the mean of the Poisson distribution, and its variance σ 2 =λ.

[0099] The observation is performed in the k-th frame, with a measurement time of (τ) and an observation data of g. k =[x k ,y k ,z k ], then z k The likelihood probability density function is: P(z) k ∨X)=Possion(z k ;λ k (X)), λ k (X) represents the average value of the radiation count.

[0100] According to the radiation propagation model, if the activity of the radioactive source is I... s The Euclidean distance between the observation point and the radiation source is D. k The local radiation count value of the environment is n b Then λ k(X) can be approximated as:

[0101] This embodiment will use g k =[x k ,y k ,z k The vectors are saved as point clouds and managed using the Farthest Point Sampling (FPS) algorithm. Assuming there are k points in the first K observations, the algorithm samples m points (m <= k). All observation points are categorized into two sets, A and B. Selected observation points are stored in set A, while unselected observation points are stored in set B. The core idea of ​​FPS is to select the point from set B that has the greatest distance to a point in set A at each observation.

[0102] Step 1, Initial situation: Set A is empty, and set B contains all point clouds.

[0103] Step 2, Select the first point: After shuffling all observation points, select the first point and add it to set A. At this point, size(A) = 1, size(B) = k-1.

[0104] Step 3, select the second point: Calculate the distance from each point in set B to a point in set A, select the point with the largest distance, and add it to set A. Here, size(A) = 2, size(B) = m - 1.

[0105] Step 4, select the third point and subsequent points: Suppose that set A has j points (j>=2) and set B has kj points. At this point, we need to define point p in set B. B The distance to a point in set A. Note that set A contains more than one point; this is the core of FPS. For p B The distance dB to set A is defined as follows:

[0106] d B =MIN

[0107] Step 4: Repeat step 4 above until set A selects m points.

[0108] like Figure 5 As shown, the particle filter algorithm mainly includes: particle initialization, weight update, particle resampling, and adaptive update.

[0109] First, in the particle initialization section, the total number of particles is set to N = 1000, represented by a set: {p i ,w i ∨i=0,1,2,……,N};where, the state of the particle is _____. Furthermore, the initial weight of each particle is 1 / N, and the initial positions of the particles are uniformly distributed in the spatial plane, satisfying the formula:

[0110] Next is the update of particle weights. After obtaining the k-th observation Zk, m valid historical observations are selected according to the previous FPS algorithm, and these m observations are used together with the current Zk to construct the joint likelihood probability density, as shown in the formula:

[0111] According to Bayesian theory, the state transition function of the particle weights is equivalent to the joint likelihood probability, thus completing the update of the particle weights.

[0112] The particle resampling process is implemented using roulette wheel (independent random sampling) to avoid particle degradation. Figure 6 As shown in the diagram. The weight of each particle corresponds to the area of ​​each award in the image. Each sampling involves rotating the disk once.

[0113] A Gaussian function is used as the adaptive update function for particles. The Gaussian distribution function is symmetric, making it easy to adjust its mean and variance; therefore, the Gaussian function is used as the particle update function. The particle parameter adjustment range differs in the initial and later stages of resampling. As the number of resampling cycles increases, the particle range gradually decreases, thus requiring adaptive adjustment of the bias parameter.

[0114] In the final stage of particle filtering, the average state of all particles is used as the predicted value for this round of filtering. The particle filtering search process is as follows: Figure 7 As shown. Finally, the particles will converge and cluster to satisfy the termination condition of the search, as... Figure 8 As shown.

[0115] Therefore, the method of this invention can be used to complete the search for unknown radioactive sources, thus effectively verifying the effectiveness of the invention. This invention introduces an FPS module based on traditional particle filtering (PF), realizing the fusion of different observation data, improving the effectiveness and robustness of the radioactive source search task. Furthermore, it can be combined with multi-robot radioactive source searches, providing a reference for multi-robot radioactive source searches.

[0116] Furthermore, it should be noted that in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element. Additionally, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. Furthermore, the character " / " in this document generally indicates an "or" relationship between the preceding and following related objects, but it may also indicate an "and / or" relationship; please refer to the context for specific interpretation. "At least one" means one or more, and "more" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can be expressed as: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0117] Furthermore, it is understood that in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0118] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0119] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of functional modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. Additionally, the functional units in the various embodiments of this invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0120] If the method is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0121] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention. It should be pointed out that although preferred embodiments of the present invention have been described, those skilled in the art, once they understand the basic inventive concept of the present invention, can make several improvements and modifications without departing from the principles described herein. These improvements and modifications should also be considered within the scope of protection of the present invention. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

Claims

1. A method for searching for radioactive sources, characterized in that, include: Construct a grid map corresponding to the search environment; Based on the grid map, radiation data from the radioactive source is observed using sensors to obtain observation data; An improved particle filter algorithm was used to process the observation data to estimate the location of the radiation source; The robot's path is navigated based on the estimated location of the radiation source and the grid map. The decision to terminate the search for the radioactive source is based on the particle distribution. The improved particle filter algorithm filters historical observation data through the farthest point sampling algorithm, merges the filtered historical observation data with the current observation data, constructs a joint likelihood probability density using the fused observation data, and updates the particle weights based on the constructed joint likelihood probability density. The process of using an improved particle filter algorithm to process the observation data to estimate the location of the radiation source includes: Define a quantity as N weighted particle set Complete particle initialization; among which, Indicates the first i The state of each particle , , Indicates the particle position. Indicates the radiation intensity of the particle; Indicates the first i The weights of each particle are assigned, with each particle's initial weight being 1 / N And a uniform distribution is used as the initial distribution of the particles; In obtaining the first k Observation data Then, the historical observation data was filtered using the farthest point sampling algorithm to select... m Valid historical observation data, based on the selected historical observation data and... Construct the joint likelihood probability density The formula is: in, Indicates probability weights; Represents the likelihood probability density of historical observations; This represents the likelihood probability density of the current observation; The particle weights are updated based on the constructed joint likelihood probability density. During particle filtering, while keeping the total number of particles constant, particles with larger weights are replicated more often, while particles with smaller weights are replicated less often. Here, particles with larger weights refer to particles with weights greater than a preset threshold, and particles with smaller weights refer to particles with weights no greater than the preset threshold. During the convergence process of particle filtering, the resampled particles are adaptively updated; the adaptive update method is to add Gaussian perturbation to the particles and distribute them according to Gaussian distribution with the original state as the expected value. In the final stage of particle filtering, the average state of all particles is used as the predicted value for this round of filtering. The method of determining whether to terminate the search for the radioactive source based on the particle distribution includes: The convergence of the particle distribution is calculated. The mean-shift algorithm is used to perform cluster analysis on the particle distribution. If the distribution radius of the particles is smaller than the preset radius value, the search is terminated; otherwise, the search continues.

2. A radioactive source searching system, characterized in that, include: The data acquisition module is used for: Construct a grid map corresponding to the search environment; Based on the grid map, radiation data from the radioactive source is observed using sensors to obtain observation data; The data processing module is used for: An improved particle filter algorithm was used to process the observation data to estimate the location of the radiation source; The robot's path is navigated based on the estimated location of the radiation source and the grid map. The decision to terminate the search for the radioactive source is based on the particle distribution. The improved particle filter algorithm filters historical observation data through the farthest point sampling algorithm, merges the filtered historical observation data with the current observation data, constructs a joint likelihood probability density using the fused observation data, and updates the particle weights based on the constructed joint likelihood probability density. The process of using an improved particle filter algorithm to process the observation data to estimate the location of the radiation source includes: Define a quantity as N weighted particle set Complete particle initialization; among which, Indicates the first i The state of each particle , , Indicates the particle position. Indicates the radiation intensity of the particle; Indicates the first i The weights of each particle are assigned, with each particle's initial weight being 1 / N And a uniform distribution is used as the initial distribution of the particles; In obtaining the first k Observation data Then, the historical observation data was filtered using the farthest point sampling algorithm to select... m Valid historical observation data, based on the selected historical observation data and... Construct the joint likelihood probability density The formula is: in, Indicates probability weights; Represents the likelihood probability density of historical observations; This represents the likelihood probability density of the current observation; The particle weights are updated based on the constructed joint likelihood probability density. During particle filtering, while keeping the total number of particles constant, particles with larger weights are replicated more often, while particles with smaller weights are replicated less often. Here, particles with larger weights refer to particles with weights greater than a preset threshold, and particles with smaller weights refer to particles with weights no greater than the preset threshold. During the convergence process of particle filtering, the resampled particles are adaptively updated; the adaptive update method is to add Gaussian perturbation to the particles and distribute them according to Gaussian distribution with the original state as the expected value. In the final stage of particle filtering, the average state of all particles is used as the predicted value for this round of filtering. The method of determining whether to terminate the search for the radioactive source based on the particle distribution includes: The convergence of the particle distribution is calculated. The mean-shift algorithm is used to perform cluster analysis on the particle distribution. If the distribution radius of the particles is smaller than the preset radius value, the search is terminated; otherwise, the search continues.

Images