A Semantic Map Matching Method and System Based on Genetic Algorithm

Abstract

The present invention belongs to the field of large-scale semantic map matching, and specifically relates to a semantic map matching method and system based on a genetic algorithm. The method includes: for the received point cloud maps A and B, determining potential matching points based on geometric and semantic distributions; constructing a transformation matrix as an individual and establishing a fitness function; relying on the fitness function to evaluate the quality of individuals, using selection, crossover, and mutation operations to guide the evolution of the population, and realizing map matching. While removing data noise, it maximally retains feature points, improving the calculation speed and accuracy; through a semi-object-level geometric feature description method, the fitness calculation between corresponding points is realized; the optimization problem of the transformation matrix is modeled as a genetic algorithm, and an adaptive variable-step mutation algorithm is proposed to jump out of local optima and achieve the maximum global optimum.

Description

Technical Field

[0001] The present invention belongs to the field of large-scale semantic map matching, and in particular relates to a semantic map matching method and system based on a genetic algorithm. Background Art

[0002] With the increasing application of robots in fields such as firefighting and rescue, and counter-terrorism, single robots are no longer able to meet mission requirements, leading to the emergence of the use of multiple robots collaborating to complete complex tasks. Compared to single robots, multi-robot collaboration can accomplish complex tasks more efficiently and robustly. However, the design of multi-robot collaborative systems is also more complex. In particular, issues such as task planning, information transmission, and multi-source heterogeneous data fusion significantly impact the stability and efficiency of multi-robot systems. The working scenes of multi-robot systems are generally large-scale or have diverse perspectives, and obtaining a global map of the scene is a fundamental issue. Limited by network bandwidth and robot computing power, a single robot is generally unable to maintain a global map of the scene. Therefore, map matching between the maps of multiple robots to establish a global map is a necessary task.

[0003] Map matching finds the transformation matrix between two maps by matching similar features between them. The maps used for fusion are generated by different robots, and the distance and viewpoint differences between the robots affect the similar features used for matching. The greater the distance between the two robots and the greater the difference in viewpoint, the greater the discrepancy in the geometric features of the same object, making matching more difficult. Therefore, algorithms (ICP / NDT) that rely solely on geometric features such as points, lines, normals, or curvature in point clouds for matching rely heavily on overlap and are significantly affected by noise. Other researchers have proposed using semantic features for fusion, achieving promising results. However, these algorithms require local maps to have a high overlap (40% or higher), a small rotation angle difference (30° or less), and a small translation error (10 meters or less). When local maps lack these conditions, or even when they do, the algorithms still suffer from significant errors. Furthermore, as map size increases, the number of point clouds increases and the features become more complex, leading to reduced accuracy and increased time. Balancing efficiency and accuracy is another challenge that matching algorithms must address. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the prior art and propose a semantic map matching method and system based on genetic algorithm.

[0005] To achieve the above objectives, the present invention proposes a semantic map matching method based on a genetic algorithm, comprising:

[0006] Step 1) For the received point cloud maps A and B, determine potential matching points based on geometric and semantic distribution;

[0007] Step 2) construct the transformation matrix into individuals and establish a fitness function;

[0008] Step 3) Rely on the fitness function to evaluate the quality of individuals, use selection, crossover and mutation operations to guide population evolution and achieve map matching.

[0009] Preferably, the step 1) comprises:

[0010] Step 1-1) Filter the point cloud maps A and B according to the semantic labels and remove points with the semantic label of road;

[0011] Step 1-2) For each point in the filtered point cloud map A, calculate the response value, mark the point with the largest response value in the neighborhood as a potential key point, compare the response value of the potential key point with the threshold ξ, and mark the point with a value greater than or equal to ξ as a key point;

[0012] Steps 1-3) Extract semantic features around key points;

[0013] Step 1-4) Determine potential matching points based on semantic features.

[0014] Preferably, the response value R in step 1-2) is:

[0015]

[0016] Among them, det(M) is the determinant of M, tr(M) represents the trace of M, and M is the point p i The covariance matrix of :

[0017]

[0018] Among them, n j For point p i The local surface normal vector of point p i The number of points in the neighborhood with a radius of δ and a center of , and the superscript T indicates transposition.

[0019] Preferably, the steps 1-3) include:

[0020] Calculate the 5 points closest to each key point and count the semantic label distribution of these 5 points

[0021] Calculate the 5 closest points around each point in point cloud map B And count the semantic label distribution of these 5 points.

[0022] Preferably, the steps 1-4) include:

[0023] For each key point in the point cloud map A Given a matrix T, the transformed coordinates If and and the number of the same labels in is greater than 3, then it is considered that and are potential matching points, where is the three-dimensional coordinate of the i-th point in the point cloud map B

[0024] Preferably, constructing the transformation matrix as an individual in step 2) includes:

[0025] Dividing the transformation matrix into the rotation angle around the Z axis and the translation vector along the XY plane, and forming an individual with this rotation angle and translation vector

[0026] Preferably, establishing the fitness function in step 2) includes:

[0027] For each potential matching point in the point cloud map A, calculate the W nearest points around it, and divide these W points into M groups according to their different semantic labels. The semantic label set and the coordinate set are respectively expressed as and where m represents the m-th group

[0028] Establish the fitness function F(T|A,B) as:

[0029]

[0030] where T is the transformation matrix and are respectively potential matching points is the point and is the probability of the corresponding point, satisfying the following formula:

[0031]

[0032] where represents and the semantic feature similarity of, when and only when and have the same semantic label, that is:

[0033]

[0034] represents and The geometric feature similarity.

[0035] Preferably, the step 3) comprises:

[0036] Step 3-1) Decompose the transformation matrix into two parts: the translation vector on the XY axis and the rotation angle around the Z axis. Encode each individual as a real number as the translation vector on the XY axis and the rotation angle around the Z axis. Generate several matrices uniformly within the upper and lower limits of the translation distance and rotation angle. Select the X with the largest fitness to form the initial population and set the initial value of the elite individual.

[0037] Step 3-2) Obtain the individual with the highest fitness in the current population and compare it with the current elite individual. If the fitness of the current individual is greater than that of the elite individual, then the current individual is the elite individual; otherwise, the elite individual is not changed.

[0038] Step 3-3) If the fitness of the elite individual is greater than the threshold or the maximum number of iterations is reached, the population evolution is terminated, otherwise go to step 3-4);

[0039] Steps 3-4) Perform roulette while retaining the best individuals in each generation;

[0040] Steps 3-5) Based on the fitness of the elite individuals, adaptively adjust the step length, perform crossover and mutation operations on the translation vector and rotation angle of each individual, and increase the number of iterations by 1;

[0041] Step 3-6) Calculate the fitness of each individual and go to step 3-2).

[0042] Preferably, the steps 3-4) include:

[0043] Step a) setting the rotation angle step length R_init = 1, the translation vector step length T_init = 2, and the maximum fitness value Threshold = 1200;

[0044] Step b) determining whether the fitness of the elite individual of the previous generation is less than Threshold / 3; if so, the rotation angle is changed to step length R_step=R_init, T_step=T_init, and executing step f); otherwise, executing step c);

[0045] Step c) determining whether the fitness of the elite individual of the previous generation is less than Threshold / 2.4. If so, the rotation angle is changed to step length R_step=R_init / 2, T_step=T_init / 2, and executing step f); otherwise, executing step d);

[0046] Step d): Determine whether the fitness of the previous generation of elite individuals is less than Threshold / 2. If the judgment is yes, the rotation angle mutation step size R_step = R_init / 5, T_step = T_init / 5, and execute step f); otherwise, execute step e).

[0047] Step e): The rotation angle mutation step size R_step = R_init / 10, T_step = T_init / 20, and execute step f).

[0048] Step f): Increment the iteration count gen by 1.

[0049] On the other hand, the present invention proposes a semantic map matching system based on a genetic algorithm, including:

[0050] A potential matching point extraction module, which is used to determine potential matching points for the received point cloud maps A and B based on geometric and semantic distributions.

[0051] A fitness calculation module based on semi-object-level features, which is used to construct the transformation matrix as an individual and establish a fitness function.

[0052] A population evolution module, which is used to evaluate the pros and cons of individuals depending on the fitness function, use selection, crossover, and mutation operations to guide the population evolution, and achieve map matching.

[0053] Compared with the prior art, the advantages of the present invention are as follows:

[0054] 1. The present invention proposes a method for calculating potential matching points based on geometric and semantic distributions, which maximally retains feature points while removing data noise, improving the calculation speed and accuracy.

[0055] 2. The present invention proposes a semi-object-level geometric feature description method, which uses the semantic and geometric distributions of some points around the sampling point to describe the features of the point and realizes the fitness calculation between corresponding points.

[0056] 3. The present invention models the optimization problem of the transformation matrix as a genetic algorithm and proposes an adaptive variable step size mutation algorithm to jump out of the local optimum and achieve the maximum global optimum. Description of the Drawings

[0057] Figure 1 is the flowchart of the semantic map matching method based on the genetic algorithm of the present invention;

[0058] Figure 2 is the algorithm registration result;

[0059] Figure 3 is the ground truth of the point cloud map;

[0060] Figure 4 is the error distribution histogram of the method of the present invention and other algorithms. Detailed implementation manners

[0061] The object of the present invention is to solve the problem of large-scale semantic map registration, and a semantic map registration method based on genetic algorithm is proposed. By extracting potential matching points, using genetic algorithm to solve the optimization matrix, and calculating fitness, the semantic map matching under large view angle differences is realized.

[0062] For local maps A and B composed of point clouds, where A = {X A , L A}, B = {X B , L B}. X A is the set of geometric coordinates of all points in the map, and there is L A is the set of semantic labels of all points in the map, and there is represents the three-dimensional coordinates of the k-th point in map A, represents the semantic label of the k-th point in map A. The goal of map registration is to estimate the transformation matrix T between two local maps, so that the error between the transformed map A and map B is the smallest, that is, the error between the estimated transformation matrix and the true value matrix is the smallest, that is:

[0063] Error(T est , T gt ) = ||log ((T gt ) -1 ·T est )|| (1)

[0064] For the convenience of description, the following definitions are made: If the point x A in map A and the point x B in map B are corresponding points, it is necessary to satisfy x A = T gt ·x B , that is, x A and x B have the same coordinates in the global map.

[0065] The semantic map matching method based on genetic algorithm models the transformation matrix as an individual, constructs a fitness function, and relies on the fitness function to guide the population evolution to complete the global optimization process. The semantic map matching method and system based on genetic algorithm include a potential matching point extraction module, a fitness calculation module based on semi-object-level features, and a population evolution module.

[0066] First, two points that may be corresponding points are extracted by the potential matching point extraction module; then, the fitness of each individual is calculated by the fitness calculation module based on semi-object-level features; finally, the population evolution module controls the direction of population evolution to find the global optimal transformation matrix.

[0067] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0068] Embodiment 1

[0069] Embodiment 1 of the present invention discloses a semantic map registration method based on a genetic algorithm.

[0070] The system flowchart is as Figure 1 shown.

[0071] 1. For the received point cloud maps A and B, based on the geometric and semantic distributions, potential matching points are determined;

[0072] Potential matching point extraction refers to finding two points that are corresponding points in maps A and B respectively, and calculating the fitness value of each individual based on the potential matching points. Its function is to find and remove the wrong points that affect the matching and improve the matching speed. The specific steps are as follows:

[0073] 1.1. Remove the points with the semantic label of road in the two local point cloud maps, because in such objects, the local features of point to point are very similar, which will make it difficult to distinguish a point according to geometric features, resulting in wrong matching.

[0074] 1.2. For each remaining point p in map A i , calculate the response value R of this point:

[0075]

[0076] where det(M) is the determinant of M, and tr(M) represents the trace of M. M is the covariance matrix of point p i :

[0077]

[0078] where n j is the local surface normal vector of this point. N is the number of points in the neighborhood with point p i as the center and a radius of δ.

[0079] Mark the point with the maximum response value in the neighborhood as the potential key point, and compare the response value of the potential key point with the threshold ξ. The points greater than or equal to ξ are marked as key points. Here, ξ = 0.01 and δ = 0.5m are taken.

[0080] 1.3. Calculate the 5 points closest to each key point and count the semantic label distribution of these 5 points At the same time, calculate the 5 points closest to each point in the local map B And count the semantic label distribution of these 5 points.

[0081] 1.4. For each key point in A, given a matrix T, the transformed coordinates If Proceed to the next step.

[0082] 1.5. If And The number of the same labels in is greater than 3, then it is considered that And Are potential matching points.

[0083] 2 Construct the transformation matrix as an individual and establish a fitness function;

[0084] In the genetic algorithm, it is necessary to calculate the fitness of each individual to evaluate the quality of the individual. In this algorithm, the transformation matrix is divided into the rotation angle around the Z-axis and the translation vector along the X and Y axes, and the rotation angle and the translation vector are combined into an individual.

[0085] Through the extraction of potential matching points, a set of potential matching points has been obtained. The fitness of the transformation matrix T will be calculated using this set of matching points. Specifically, the probability that two points are corresponding points under the transformation matrix T is calculated. The larger the probability value, the more likely the two points are corresponding points and the closer the transformation matrix T is to the true value matrix. The geometric and semantic distribution characteristics of N surrounding points are used to calculate the probability that two points are the same point.

[0086] For each point, the W closest points around it are calculated. These W points are divided into M groups according to their different semantic labels. The semantic label set and the coordinate set are respectively represented as And Among them,

[0087] Specifically, given the matrix T, the points And The probability that they are corresponding points is expressed as:

[0088]

[0089] Represents And The semantic feature similarity of. When and only when And Have the same semantic label, That is:

[0090]

[0091] express and The geometric feature similarity measures the geometric distance between two points.

[0092]

[0093] in express and The residual, for res k,i The covariance matrix of .

[0094]

[0095] R is the rotation matrix, For point The covariance matrix in the corresponding half-object level represents the distance from map A to The geometric coordinate covariance of the point with semantic label M among the nearest W points. Since the point with semantic label M is part of the object of category M, not the whole object, it is called the semi-object level. It reflects the point Spatial distribution of nearby objects.

[0096] Therefore, for the transformation matrix T, its fitness is expressed as:

[0097]

[0098] 3. Rely on the fitness function to evaluate the quality of individuals, use selection, crossover and mutation operations to guide population evolution, and achieve map matching.

[0099] Its role is to control the direction of population evolution, making the population evolve in a direction with smaller errors, avoid falling into local optimality, and maintain population diversity. It includes steps such as population initialization, fitness calculation, selection, crossover and mutation, and termination condition determination.

[0100] 3.1: Population Initialization: Decompose the transformation matrix into two components: the translation vector on the X and Y axes and the rotation angle around the Z axis. Encode each individual as a real number representing the translation vector on the X and Y axes and the rotation angle around the Z axis. The translation distance must be within the upper and lower limits of -15 meters to 15 meters, and the rotation angle must be within the upper and lower limits of 0 to 180 degrees.

[0101] Since the genetic algorithm is sensitive to initial values, 2000 matrices are first generated uniformly within the upper and lower limits of the translation distance and rotation angle, and the 100 matrices with the largest fitness are selected as the first generation population.

[0102] 3.2: Fitness calculation: The fitness of all individuals is calculated.

[0103] 3.3: Selection: The roulette wheel algorithm is used for selection, and the optimal individual of each generation is retained.

[0104] 3.4: Crossover and mutation: The translation vector and rotation angle of each individual are respectively subjected to crossover and mutation operations. Real number step size crossover is performed with a specific probability, and random step sizes are added to the translation vector and rotation angle with a specific probability. An adaptive variable step size mutation algorithm is designed. Based on the fitness of the elite individuals, the mutation step size is adaptively adjusted.

[0105] Specifically, when the fitness value of the elite individual is less than a certain threshold, the mutation step size is changed. The specific steps are as follows:

[0106] a) Initialization: The initial rotation angle mutation step size R_init = 1, the initial translation vector mutation step size T_init = 2, and the maximum fitness value Threshold = 1200;

[0107] b) If the fitness fitness of the elite individual in the previous generation is less than Threshold / 3, the rotation angle mutation step size R_step = R_init, T_step = T_init; if it is greater than or equal to Threshold / 3, then step c) is executed;

[0108] c) If the fitness fitness of the elite individual in the previous generation is less than Threshold / 2.4, the rotation angle mutation step size R_step = R_init / 2, T_step = T_init / 2; if it is greater than or equal to Threshold / 2.4, then step d) is executed;

[0109] d) If the fitness fitness of the elite individual in the previous generation is less than Threshold / 2, the rotation angle mutation step size R_step = R_init / 5, T_step = T_init / 5; if it is greater than or equal to Threshold / 2, then step e) is executed;

[0110] e) The rotation angle mutation step size R_step = R_init / 10, T_step = T_init / 20.

[0111] 3.5: Termination condition judgment: The evolution is terminated when the maximum number of iterations is reached or the fitness value of the elite individual reaches the threshold, otherwise steps 2)-4) are repeated.

[0112] Embodiment 2

[0113] Embodiment 2 of the present invention discloses a semantic map matching system based on a genetic algorithm, which is implemented by the method of Embodiment 1. The system includes:

[0114] A potential matching point extraction module, which is used to determine potential matching points for the received point cloud maps A and B based on geometric and semantic distributions;

[0115] A fitness calculation module based on semi-object-level features, which is used to construct a transformation matrix as an individual and establish a fitness function;

[0116] A population evolution module, which is used to evaluate the quality of individuals relying on the fitness function, use selection, crossover, and mutation operations to guide the population evolution, and achieve map matching.

[0117] Verification experiment

[0118] To verify the performance of the proposed algorithm, experiments were carried out on the dataset, and some algorithms were selected for comparison. Experiments were carried out on sequence 05 of the KITTI dataset. The point cloud map was established using LIO-SLAM, and then the point cloud map was segmented into two local sub-maps with an overlap rate of 30%. One of the local sub-maps was multiplied by a rotation matrix to form map A. The other local sub-map formed map B.

[0119] Taking the example of translating -15 meters in both the X and Y axes and rotating 60° around the Z axis, that is, the corresponding point coordinates of the local maps A and B differ by 15 meters in the X and Y axis directions and are rotated 60° around the Z axis. These two maps are sent into the matching algorithm of this paper, and the matching results are as Figure 2 shown.

[0120] Among them, red represents the local map B, and blue represents the rotated local A. It can be seen that the two sub-maps are very well registered, and the error of the transformation matrix is 0.00010. It is very similar to the ground truth of the point cloud map.

[0121] Figure 3 is the ground truth of the point cloud map.

[0122] Furthermore, the fusion error was tested under the conditions that the rotation angle ranges from 0 to 180° with a step of 3°, and the translation ranges from -15 meters to 15 meters with a step of 3 meters. And two algorithms, lcr and VPFBR, were selected as comparison algorithms. The evaluation index is the error between the rotation matrix output by the algorithm and the ground truth rotation matrix. The error distribution histograms of the three algorithms are as Figure 4 shown. It can be seen that the algorithm errors of this paper's algorithm are mostly around 0, and at the point where the error is 0, the number of the algorithm proposed in this paper is the largest.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the embodiments, those of ordinary skill in the art should understand that any modification or equivalent replacement of the technical solutions of the present invention does not depart from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A semantic map matching method based on genetic algorithm, comprising: Step 1): For the received point cloud maps A and B, determine potential matching points based on geometric and semantic distributions; Step 2): Construct the transformation matrix as an individual and establish a fitness function; Step 3): Rely on the fitness function to evaluate the quality of individuals, use selection, crossover and mutation operations to guide the evolution of the population, and achieve map matching; The said Step 1) includes: Step 1-1): Filter the point cloud maps A and B according to semantic labels, and remove the points with the semantic label of road; Step 1-2): For each point in the filtered point cloud map A, calculate the response value respectively, mark the point with the largest response value in the neighborhood as a potential key point, compare the response value of the potential key point with the threshold ξ, and mark the points greater than or equal to ξ as key points; Step 1-3): Extract the semantic features around the key points; Step 1-4): Determine potential matching points according to the semantic features.

2. The semantic map matching method based on genetic algorithm according to claim 1, characterized in that, The response value R in the said Step 1-2) is: where det(M) is the determinant of M, tr(M) represents the trace of M, and M is the covariance matrix of point p i : where n j is the local surface normal vector of point p i , N is the number of points within a neighborhood centered at point p i with a radius of δ, and the superscript T represents the transpose.

3. The semantic map matching method based on genetic algorithm according to claim 1, characterized in that The said Step 1-3) includes: Calculate the 5 points closest to each key point and count the semantic label distribution of these 5 points Calculate the 5 points closest to each point in the point cloud map B And count the semantic label distribution of these 5 points.

4. The semantic map matching method based on genetic algorithm according to claim 1, characterized in that The said Step 1-4) includes: For each key point in the point cloud map A Given a matrix T, the transformed coordinates If and and If the number of the same labels in and is greater than 3, then and are potential matching points, where is the three-dimensional coordinates of the i-th point in the point cloud map B.

5. The semantic map matching method based on genetic algorithm according to claim 1, characterized in that In the said Step 2), constructing the transformation matrix as an individual includes: Divide the transformation matrix into the rotation angle around the Z axis and the translation vector along the XY plane, and form an individual with this rotation angle and translation vector.

6. The semantic map matching method based on genetic algorithm according to claim 1, characterized in that In the said Step 2), establishing the fitness function includes: For each potential matching point in the point cloud map A, calculate the W nearest points around it, and divide these W points into M groups according to their different semantic labels. The semantic label set and the coordinate set are respectively expressed as and where m represents the m-th group; Establish the fitness function F(T|A,B) as: where T is the transformation matrix, and are potential matching points, is the point and is the probability of the corresponding points, satisfying the following formula: Among them, represents and the semantic feature similarity of, if and only if and have the same semantic label, that is: representation and geometric feature similarity.

7. The semantic map matching method based on genetic algorithm according to claim 1, wherein The said Step 3) includes: Step 3-1): Decompose the transformation matrix into two parts: the translation vector on the XY axis and the rotation angle around the Z axis, encode each individual real number as the translation vector on the XY axis and the rotation angle around the Z axis, uniformly generate several matrices within the upper and lower limits of the translation distance and rotation angle, select the X individuals with the largest fitness to form the initial population, and set the initial value of the elite individual; Step 3-2): Obtain the individual with the largest fitness in the current population, compare it with the current elite individual. If the fitness of the current individual is greater than the fitness of the elite individual, the current individual is the elite individual; otherwise, the elite individual remains unchanged; Step 3-3): If the fitness of the elite individual is greater than the threshold or reaches the maximum number of iterations, terminate the population evolution; otherwise, go to Step 3-4); Step 3-4): Perform roulette wheel selection, and retain the best individual of each generation; Step 3-5): Based on the fitness of the elite individual, adaptively adjust the mutation step size, perform crossover and mutation operations on the translation vector and rotation angle of each individual respectively, and increase the number of iterations by 1; Step 3-6): Calculate the fitness of each individual, and go to Step 3-2).

8. The semantic map matching method based on genetic algorithm according to claim 7, characterized in that The said Step 3-4) includes: Step a): Set the rotation angle mutation step size R_init = 1, the translation vector mutation step size T_init = 2, and the maximum fitness value Threshold = 1200; Step b): Judge whether the fitness fitness of the elite individual in the previous generation is less than Threshold / 3. If it is judged to be yes, the rotation angle mutation step size R_step = R_init, T_step = T_init, and execute Step f); otherwise, execute Step c); Step c): Determine whether the fitness of the previous generation of elite individuals is less than Threshold / 2.

4. If the determination result is yes, set the rotation angle mutation step size R_step = R_init / 2 and T_step = T_init / 2, and execute step f); otherwise, execute step d). Step d): Determine whether the fitness of the previous generation of elite individuals is less than Threshold / 2. If the determination result is yes, set the rotation angle mutation step size R_step = R_init / 5 and T_step = T_init / 5, and execute step f); otherwise, execute step e). Step e): Set the rotation angle mutation step size R_step = R_init / 10 and T_step = T_init / 20, and execute step f). Step f): Increment the iteration count gen by 1.

9. A system for the semantic map matching method based on the genetic algorithm according to claim 1, characterized in that It includes: A potential matching point extraction module, which is used to determine potential matching points for the received point cloud maps A and B based on geometric and semantic distributions. A fitness calculation module based on semi-object-level features, which is used to construct a transformation matrix as an individual and establish a fitness function. and A population evolution module, which is used to evaluate the quality of individuals depending on the fitness function, use selection, crossover, and mutation operations to guide the population evolution, and achieve map matching.

Images