A terminal positioning method based on dynamic particle filtering of map information

Abstract

The application discloses a terminal positioning method based on dynamic particle filtering of map information, and aims at the problem that a common particle filtering algorithm is difficult to determine a good initial particle number and distribution state in the technical field of indoor positioning technology. The indoor positioning vector map is made according to a self-defined map format, the best reference particle number is obtained by field measurement of different areas, the particle distribution proportion of each area is calculated according to the received RSSI data, the final particle number and particle distribution scheme are further calculated and determined, the particles are initialized according to the particle distribution scheme, the position estimation is obtained through the particle filtering positioning algorithm, and the positioning precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of indoor positioning technology, and particularly relates to a terminal positioning method based on dynamic particle filtering of map information. BACKGROUND

[0002] In the indoor positioning scenario, the particle filtering algorithm is commonly used to fuse multi-dimensional information to improve positioning accuracy, which generally includes the following five steps: initialization of particles, particle transfer, weight update, resampling and position estimation. Among them, two problems need to be solved in the particle initialization stage, one is the number of particles, and the other is the distribution of particles.

[0003] Firstly, the number of particles has a great influence on the effect and calculation amount of particle filtering. Generally, the more the number of particles, the more the sampling of distribution, and the more accurate the representation of the distribution of state space, so as to obtain good positioning effect. However, with the increase of the number of particles, the calculation amount is also large, because each particle will participate in the state transfer and weight calculation. In the indoor positioning scenario, there is a requirement for real-time positioning, so the positioning delay cannot be too high. In the case of too many particles, the server will spend a lot of time to continue calculation, resulting in the increase of the delay of obtaining positioning results. Therefore, when determining the number of particles, the influence of the number of particles on the positioning accuracy and calculation efficiency should be considered comprehensively, so as to select a reasonable number of particles.

[0004] Meanwhile, for the distribution of particles, the distribution of particles is usually determined according to prior knowledge in the initialization stage. For the positioning scenario, the particle distribution area needs to be determined before the initialization of particles, and then the particles are placed in the area. The size of the particle distribution area affects the final positioning effect. When the area is large, for example, the area is limited to the whole floor, the distribution of particles is very sparse, which will be difficult to cover comprehensively, and at the same time, a large number of particles are far away from the real position, so that the calculated weight is too small. After a round of filtering, only a small number of particles survive, and the position estimation calculated by these particles often has a large error. When the area is small, the distribution of particles is too compact, and the discrimination degree of each particle is not enough, resulting in the decline of the final positioning effect.

[0005] Therefore, different number of particles and particle distribution have a great influence on the positioning accuracy and positioning delay of the particle filtering method in the indoor positioning scenario. The terminal positioning method based on dynamic particle filtering of map information proposed in the present application combines map information and beacon deployment information, and gives a more reasonable particle number and distribution scheme in the particle initialization stage, which improves the performance of the overall particle filtering positioning method. SUMMARY

[0006] To solve the problems in the background art, the application provides a terminal positioning method based on dynamic particle filtering of map information, so as to give a more reasonable particle number and distribution scheme in the particle initialization stage and improve the performance of the overall particle filtering positioning method.

[0007] To achieve the above object, the application provides the following technical scheme.

[0008] A terminal positioning method based on dynamic particle filtering of map information, comprising:

[0009] S1: drawing a map in SVG format with multiple areas according to architectural drawings;

[0010] S2: a test person measures in the field with a positioning terminal, sets the reference particle number in each area according to the area type, complexity and size of different areas in the map, and one particle corresponds to one feasible position of the positioning terminal on the map;

[0011] S3: deploying at least one beacon in each area in the field and deploying virtual beacons on the map corresponding to the number and position of the beacons in the field, the positioning terminal is used to receive the received RSSI sent by the beacons in the field, and the RSSI after normalization processing is calculated according to the value size, and the proportion of the RSSI corresponding to the beacon is obtained;

[0012] S4: calculating the particle distribution proportion of each area according to the proportion of the beacons, and determining the final particle number and particle distribution scheme according to the particle distribution proportion;

[0013] S5: particle initialization according to the final particle number and particle distribution scheme;

[0014] S6: moving the position of each particle while the map wall information is assisted to perform particle collision detection;

[0015] S7: after the particle moves, the positioning terminal receives the RSSI again, and updates the weight of the particle according to the RSSI received again;

[0016] S8: resampling all particles;

[0017] S9: weighted average of the particle set after resampling can obtain the positioning information of the positioning terminal;

[0018] S10: repeating S5 to S9 can continuously track the position information of the positioning terminal.

[0019] Preferably, the definition of the area in S1 is that: each space including a corridor and a room in the map is represented as an SVG geometric vector graph, each geometric vector graph corresponds to a unique area, and the set of all areas in the entire map is represented as A={a1,...,a i}。

[0020] Preferably, the definition of the area type in S2 is that: the area type is determined according to the complexity of the geometric vector graph and the size of the area The area type belongs to, the set of area types is represented as , and one area type corresponds to multiple areas.

[0021] Preferably, in step 4, each area type The set of corresponding reference particle numbers can be represented as Each beacon is bound to an area type t j , and the beacon is used to send RSSI to the positioning terminal at the particle position. The set of beacons is represented as The set of intensity indications RSSI sent by the beacon is represented as The set of proportions of the intensity indications RSSI sent by each beacon is represented as For each beacon , the corresponding area is queried, the area type is queried, the reference particle number corresponding to the area type is queried, and the number of particles that should be allocated to the area is calculated according to the formula Finally, the allocated particle numbers of all areas are obtained .

[0022] Preferably, in S5, the number of particles is initialized in each area in a Gaussian distribution.

[0023] Preferably, in S7, all particles in the area calculate the distances between themselves and the virtual beacons through RSSI, and assign high and low weights to the particles according to the similarity of the calculated distances and the measured distances by comparing the similarity.

[0024] Compared with the prior art, the beneficial effects of the present application are:

[0025] 1. The initial distribution of the particles considers the positions of the beacons sending signals, narrows the range of the initial distribution of the particles, and retains as many effective particles as possible.

[0026] 2. Based on the idea of ​​map region division, the number of reference particles is different in different regions, which avoids the waste of computing resources caused by using a fixed number of particles.

[0027] 3. By considering the specific circumstances of RSSI, ensure that the number of particles allocated to each region is within a reasonable range, avoiding excessive concentration or dispersion of particles.

[0028] 4. Since the number of particles varies in different positioning areas, resulting in different positioning accuracies, this invention can effectively determine the optimal number of particles based on the area type, thereby improving positioning accuracy. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention;

[0030] Figure 2 This is a flowchart illustrating the specific algorithm of the present invention;

[0031] Figure 3 This is a pseudocode diagram of the particle number allocation part of the present invention;

[0032] Figure 4 A graph showing the relationship between the number of particles and the average positioning error in different regions;

[0033] Figure 5 The cumulative distribution function of positioning error for the proposed dynamic particle number allocation method and fixed particle number method is shown in the figure.

[0034] Figure 6 A comparison of computation time between the proposed dynamic particle number allocation method and the fixed particle number method. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example 1

[0037] like Figure 1 As shown, a terminal localization method based on dynamic particle filtering of map information includes:

[0038] S1: Based on the architectural drawings, rooms, doors, corridors, and other spaces are uniformly represented as scalable vector graphics. Each vector graphic can also contain attribute data such as name and floor. After completion, an SVG format map containing the architectural structure and layout is obtained. Finally, after being parsed by the program, the interior structure is saved line by line to the database.

[0039] S2: Testers use positioning terminals to measure in the field. Based on the area type, complexity and size of different areas in the map, the number of reference particles in each area is set. One particle corresponds to a feasible location of the positioning terminal on the map.

[0040] S3: Deploy one beacon in each area of ​​the field and deploy virtual beacons on the map corresponding to the number and location of the beacons in the field. The positioning terminal is used to receive the RSSI signal sent by the beacons in the field, normalize the received RSSI, and calculate the relative proportion of the normalized RSSI according to the value to obtain the proportion of beacons corresponding to the RSSI.

[0041] S4: Calculate the particle allocation ratio for each region based on the proportion occupied by the beacon, and then calculate and determine the final number of particles and particle distribution scheme based on the particle allocation ratio.

[0042] S5: Initialize particles based on the final number of particles and particle distribution scheme;

[0043] S6: Move the position of each particle, and use the map wall information as an aid to perform particle collision detection;

[0044] S7: After the particle completes its movement, the positioning terminal receives the RSSI again and updates the particle's weight based on the received RSSI.

[0045] S8: Resample all particles to ensure that high-weight particles survive and eliminate low-weight particles.

[0046] S9: The positioning information of the positioning terminal can be obtained by weighting the set of resampled particles;

[0047] S10: Repeat S5 to S9 to continuously track and locate the position information of the terminal.

[0048] In this embodiment, the initial distribution of particles takes into account the beacon position of the transmitted signal, which narrows the range of the initial particle distribution and preserves as many effective particles as possible. Furthermore, based on the idea of ​​map region division, the number of reference particles in different regions is different, avoiding the waste of computing resources caused by using a fixed number of particles. Then, by considering the specific situation of RSSI, it is ensured that the number of particles allocated to each region is reasonable, avoiding particles being too concentrated or too dispersed. Finally, this invention can effectively determine the optimal number of particles according to the region type, thereby improving positioning accuracy.

[0049] Example 2

[0050] The difference between this embodiment and Embodiment 1 is that: Figure 2 As shown, the specific algorithm flow of this application is as follows:

[0051] 1. At the beginning of this algorithm, the parameters are initialized. Here, C is the set of particle numbers corresponding to each region, and each... The initial value is 0; S is the proportion of the corresponding beacon calculated based on RSSI; N is the number of reference particles corresponding to each region type, for example, Figure 4 The curves showing the relationship between particle number and average positioning error in three different regions demonstrate how the optimal particle number for each region can be determined through field measurements. Figure 4 In the example, the number of reference particles in region 1 is 100, the number of reference particles in region 2 is 150, and the number of reference particles in region 3 is 300; A is the set of all regions in the entire map; B is the set of all beacons in the positioning environment.

[0052] 2. Next, create a vector map, dividing the area into regions using geographical spaces such as rooms, doors, and corridors;

[0053] 3. Deploy beacons in the environment and map their distribution information to different areas of the map;

[0054] 4. When the positioning terminal moves in the positioning environment, it receives several RSSIs. After filtering the RSSIs, the set of RSSIs at that moment is obtained, denoted as: ;

[0055] 5. According to Calculate the corresponding normalized value to obtain Then calculate the proportion of each RSSI to the total data, expressed as The larger the value of 's', the closer the distance to the beacon.

[0056] Example 3

[0057] The difference between this embodiment and Embodiment 2 is that, as Figure 3 As shown, Figure 3 This is a pseudocode diagram of the particle number allocation part based on map region division in the present invention, which includes the following steps:

[0058] Step 1: Iterate through R, querying the corresponding beacon b and the corresponding relative proportion. and the area to which the beacon belongs. ;

[0059] Step 2: Based on the region Get the number of particles currently allocated to this region. According to the region Obtain the number of reference particles in this type of region. The number of particles allocated is obtained. Therefore, the current number of particles allocated to this region is... ;

[0060] Step 3: judging whether R is completed, if not, jumping to step 1 to continue the iteration, if yes, continuing to the next step;

[0061] Step 4: obtaining the distribution of particles in all regions after the iteration , and obtaining reasonable initial particle number and particle distribution according to C.

[0062] In the embodiment, as shown in Figure 5 , Figure 5 is a cumulative error function diagram of the method proposed by the present application and the traditional particle filter method with fixed particle number, and it can be seen that the proposed method is superior to the traditional method in overall accuracy, as shown in Figure 6 , Figure 6 is a calculation time comparison diagram of the method proposed by the present application and the traditional particle filter method with fixed particle number, and it can be seen that the proposed method can save an average of 34 ms of calculation time in single positioning.

[0063] Those skilled in the art will realize that the embodiments described herein are for the purpose of helping the reader understand the principles of the present application, and should be understood as not limiting the scope of protection of the present application to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations according to the technical inspiration disclosed by the present application without departing from the essence of the present application, and these modifications and combinations are still within the scope of protection of the present application.

Claims

1. A terminal positioning method based on map information dynamic particle filtering, characterized in that, The method comprises the following steps: S1: draw a map in SVG format with multiple areas according to the building drawings; S2: a tester holds a positioning terminal to measure in the field, sets the number of reference particles in each area according to the area type, complexity and size of different areas in the map, and one particle corresponds to one feasible position of the positioning terminal on the map; S3: deploy at least one beacon in each area in the field, and deploy virtual beacons corresponding to the number and position of the beacons in the field on the map, the positioning terminal is used to receive the RSSI sent by the beacons in the field, normalize the received RSSI, and calculate the relative proportion of the normalized RSSI according to the numerical value, to obtain the proportion of the RSSI corresponding to the beacon; S4: calculate the particle distribution proportion of each area according to the proportion of the beacons, and determine the final particle number and particle distribution scheme according to the particle distribution proportion; S5: initialize the particles according to the final particle number and particle distribution scheme; S6: move the position of each particle while the map wall information is assisted to perform particle collision detection; S7: after the particles are moved, the positioning terminal receives the RSSI again, and updates the weight of the particles according to the RSSI received again; S8: resample all particles; S9: weighted average of the particle set after resampling can obtain the positioning information of the positioning terminal; S10: repeat S5 to S9 to continuously track the position information of the positioning terminal.

2. The terminal positioning method based on map information and dynamic particle filtering according to claim 1, wherein, The definition of the regions in S1 is: each space including a corridor and a room in the map is represented as an SVG geometric vector graphics, each geometric vector graphics corresponds to a unique region, and the set of all regions in the entire map is represented as A={a1,...,a i}.

3. The terminal positioning method based on map information and dynamic particle filtering according to claim 2, characterized in that, The definition of the region type in S2 is: according to the complexity of the geometric vector figure and the size of the region to determine the region The region type set is represented as , and one region type corresponds to multiple regions.

4. The terminal positioning method based on map information and dynamic particle filtering according to claim 3, characterized in that, In step 4, each region type The set of corresponding reference particle numbers can be represented as Each beacon is associated with a region type t j Binding, beacons are used to send RSSI to the positioning terminal at the particle's location, and the set of beacons is represented as beacon The set of transmitted strength indicators (RSSIs) is represented as follows: Each beacon The set of proportions of transmitted strength indicators (RSSI) is represented as follows: For each beacon Query its corresponding region Region type The region type corresponds to the number of reference particles. The region is calculated according to the formula. Number of particles to be allocated Finally, the number of particles allocated to all regions is obtained. .

5. The method of claim 4, wherein the map information is obtained from a map server. Initialization of the particles in each region in S5 with a Gaussian distribution particles.

6. The method of claim 1, wherein, In S7, all particles in the area calculate the distance between themselves and each virtual beacon through RSSI, compare the similarity of the calculated distance and the measured distance, and assign high and low weights to each particle according to the similarity of the distance.

Images