Training method of model for positioning, positioning method, electronic equipment and medium

Abstract

The invention relates to the technical field of electronic equipment, in particular to a training method of a model for positioning, a positioning method, electronic equipment and a medium. The method comprises the following steps: acquiring a plurality of pieces of visible light data collected at a plurality of positions under different ambient lights at each position; and taking the visible light data, the position information corresponding to the visible light data and the ambient light identifier of the ambient light corresponding to the visible light data as training data samples, and inputting the training data samples into the first neural network model to train the first neural network model to obtain a second neural network model. Therefore, according to the model training method for positioning provided by the embodiment of the invention, the obtained second neural network model can be applied to positioning under various complex ambient light interferences. Moreover, the second neural network model is obtained by training the visible light data under different ambient lights, so that the influence of the ambient lights can be inhibited, and the positioning precision is higher.

Description

Training method for positioning model, positioning method, electronic device and medium Technical Field

[0001] The present application relates to the technical field of electronic equipment, and in particular to a training method for a positioning model, a positioning method, an electronic device, and a medium. Background Art

[0002] With the development of positioning technology, electronic devices can meet the positioning needs of various application scenarios, such as indoors and underground. For positioning scenarios such as indoors and underground where satellite signals are weak, visible light positioning technology based on light-emitting diodes (LEDs) can be used to achieve both positioning accuracy and cost. Current visible light positioning technology generally uses the Lambertian model to model the mapping relationship between light intensity and distance, thereby estimating the location of the target receiving device.

[0003] For example, the receiving device 100 shown in FIG1 is located in an indoor space 200. Indoor space 200 includes five LED lights: LED1 to LED5, and a three-dimensional spatial coordinate system can be established based on indoor space 200. Receiving device 100 can receive the transmission signals of the five LED lights at its current position (x, y, z), process the received information, obtain the strengths of the five transmission signals, and calculate the current position coordinates of the receiving device based on the strengths of the transmission signals and the Lambertian model.

[0004] However, the Lambertian model in the current visible light positioning technology based on light intensity does not consider the impact of ambient light, noise, different LED lights, etc. on the positioning results, and the positioning accuracy is low.

[0005] Summary of the Invention

[0006] The embodiments of the present application provide a training method for a positioning model, a positioning method, an electronic device, and a medium, which solve the technical problem that the Lambertian model under different calibration ambient light conditions in visible light-based positioning technology cannot be universally used.

[0007] In a first aspect, an embodiment of the present application provides a training method for a model for positioning, comprising: obtaining a plurality of visible light data collected at a plurality of positions under different ambient light conditions at each position, wherein each visible light data comprises light data from a different visible light source; taking each visible light data, the position information corresponding to each visible light data, and the ambient light identification of the ambient light corresponding to each visible light data as training data samples, and inputting them into a first neural network model to train the first neural network model and obtain a second neural network model.

[0008] It can be understood that the first neural network model is the neural network model in the following embodiments, and the second neural network model is the positioning model in the following embodiments.

[0009] It can be understood that each visible light data may be mixed light data of all visible light sources received at a certain moment and at a certain position.

[0010] It can be understood that the parameters in the first neural network model are initialized parameters. The training data samples are input into the first neural network model for training. The parameters of the first neural network model can be optimized based on the training data samples to obtain a second neural network model. When optimizing the parameters of the first neural network model, the first neural network model can learn the impact of different ambient light on visible light data, learn the relationship between visible light data at the same location, and learn the correspondence between the feature data of the visible light data and the location information. The relationship between the visible light data and the corresponding location is modeled to obtain a second neural network model that can suppress the impact of ambient light on the positioning results and has higher positioning accuracy.

[0011] The positioning model training method provided in the embodiments of the present application collects visible light data under different ambient light conditions to train a first neural network model. The resulting second neural network model can be applied to positioning under various complex ambient light interference conditions. Furthermore, because the second neural network model is trained using visible light data under different ambient light conditions, it can suppress the influence of ambient light and achieve higher positioning accuracy.

[0012] It is understood that the second neural network model in the embodiments of the present application can be applied to positioning scenarios that include multiple visible light sources, including but not limited to environments where satellite signals are weak, such as underground, indoors, and outdoors with many obstructions. The visible light sources can include LED lights, energy-saving lamps, incandescent lamps, and other light sources that can emit visible light signals.

[0013] In a possible implementation of the first aspect above, the position information corresponding to the visible light data is the coordinates of the position of the visible light data obtained by taking a preset position in the positioning scene as the origin of a spatial coordinate system.

[0014] In a possible implementation of the first aspect above, a plurality of visible light data under different ambient light at each position collected at multiple positions is obtained, including: obtaining the visible light data at each position collected at multiple positions under preset ambient light; and performing data enhancement on the visible light data to obtain a plurality of visible light data under different ambient light at each position.

[0015] It can be understood that the influence of different ambient light on visible light sources is similar and obeys the Lambert model, but there are inconsistencies in the scale of the Lambert model. By enhancing the visible light data under a certain ambient light, for example, multiplying the RSS data by a coefficient, the RSS data under other ambient light environments can be obtained. There is no need to set up a data collection environment for multiple ambient light environments. By simply enhancing the visible light data under a certain ambient light, visible light data under different ambient light environments can be obtained, saving time and effort. In addition, based on the neural network learning method, the relationship between different ambient light and positioning results can be learned, which can eliminate the influence of ambient light on positioning results and obtain a high-precision second neural network model that is independent of ambient light.

[0016] In a possible implementation of the first aspect above, each visible light data, the position information corresponding to each visible light data, and the ambient light identification of the ambient light corresponding to each visible light data are used as training data samples, including: separating the light data corresponding to different visible light sources in the visible light data to obtain a first RSS data sample corresponding to each visible light data, the first RSS data sample including multiple first RSS data corresponding to different visible light sources; each first RSS data sample, the position information corresponding to each first RSS data sample, and the ambient light identification corresponding to each first RSS data sample are used as training data samples.

[0017] In a possible implementation of the first aspect above, the first neural network model includes a first neural network sub-module and a second neural network sub-module; each visible light data, the position information corresponding to each visible light data, and the ambient light identification of the ambient light corresponding to each visible light data are input as training data samples into the first neural network model, including: each first RSS data sample, the position information corresponding to each first RSS data sample, and the ambient light identification corresponding to each first RSS data sample are input as training data samples into the first neural network sub-module, wherein the first neural network sub-module takes RSS data from different visible light sources under any ambient light at any position as input, and takes weights corresponding to each RSS data as output; based on the weights corresponding to each RSS data in the first RSS data sample output by the first neural network sub-module, the RSS data in the first RSS data sample is weighted to obtain a second RSS data sample; each second RSS data sample, the position information corresponding to each second RSS data sample, and the ambient light identification corresponding to each second RSS data sample are input into the second neural network sub-module.

[0018] It can be understood that different visible light sources have different distances from the position of the visible light data to be located, so the signal-to-noise ratio during the transmission of visible light data in the environment is also different, and the visible light data is affected differently by the noise in the ambient light. The first neural network sub-module learns the mapping relationship between the light data of different visible light sources received, the calculated distance between the visible light source and the position of the received data, and the error between the actual distance, and obtains the weights of the light data of different visible light sources based on the visible light data, so that the second neural network sub-module is more dependent on the light data of the visible light source that is closer to the position where the visible light data is received during training, which can reduce the influence of noise on the positioning results of the second neural network model and improve the positioning accuracy.

[0019] In a possible implementation of the first aspect above, the first neural network submodule includes an attention mechanism.

[0020] In a possible implementation of the first aspect above, the first neural network submodule is a multi-layer perceptron with an attention mechanism.

[0021] In a possible implementation of the first aspect above, the second neural network submodule includes a long short-term memory network layer and a fully connected layer.

[0022] In a possible implementation of the first aspect above, the method further includes: optimizing parameters of the second neural network model based on the predicted positioning data output by the second neural network model and the actual position data corresponding to the predicted positioning data.

[0023] In a possible implementation of the first aspect above, optimizing the parameters of the second neural network model includes: based on the predicted positioning data and the actual position data, using a loss function to determine the error between the predicted positioning data and the actual position data; and optimizing the parameters of the second neural network model based on the error.

[0024] In a possible implementation of the first aspect, the loss function is expressed by the following formula:

[0025]

[0026] Among them, β represents the regularization adjustment coefficient of the regularization term; N represents the number of visible light sources; d j Represents the distance between the position corresponding to the visible light data calculated based on the Lambert model and the jth visible light source; RSS j represents the light data corresponding to the jth visible light source; ε represents the channel noise, and the channel noise is Gaussian noise.

[0027] It is understandable that the traditional loss function may cause the second neural network model to overfit, which in turn will affect the stability of the second neural network model. By backpropagating the loss function with the regularization term added and optimizing the parameters of the second neural network model, the anti-interference ability (i.e., robustness) of the second neural network model can be enhanced.

[0028] In a possible implementation of the first aspect, at least some of the multiple positions correspond to the same motion trajectory.

[0029] It can be understood that in some embodiments, the visible light data in the training data sample can be the visible light data of the device at multiple positions in different motion trajectories, and the training data can include visible light data, the time corresponding to the visible light data, the position corresponding to the visible light data, etc. Furthermore, the first neural network model can include a neural network model that can process time series data and spatial data, such as a recurrent neural network model (RNN), LSTM, etc. Such a neural network model can not only learn the impact of different ambient light on visible light data, the impact between different visible light light sources, and the correspondence between visible light data and position, but also learn the relationship between visible light data corresponding to continuous positions on a continuous motion trajectory, for example, the relationship between the continuous changes in visible light data at adjacent positions. The positioning model can also use the learned relationship between visible light data corresponding to continuous positions as a constraint condition for the distance between the visible light source and the device. Furthermore, the positioning accuracy of the trained second neural network model is higher.

[0030] In other embodiments, the visible light data in the training data sample may be visible light data of any discrete multiple positions in the positioning scene.

[0031] In a second aspect, an embodiment of the present application provides a training device for a positioning model, comprising: a data acquisition module for acquiring a plurality of visible light data collected at a plurality of locations under different ambient light conditions at each location, wherein each visible light data includes light data from a different visible light source;

[0032] The model training module is used to input each visible light data, the position information corresponding to each visible light data, and the ambient light identification of the ambient light corresponding to each visible light data as training data samples into the first neural network model to train the first neural network model and obtain a second neural network model.

[0033] In a third aspect, an embodiment of the present application provides a positioning method, including: obtaining collected measured visible light data; inputting the measured visible light data into a second neural network model trained by the training method of the model for positioning in the first aspect above, and outputting a predicted position corresponding to the measured visible light data.

[0034] In a fourth aspect, an embodiment of the present application provides a positioning device, comprising: a data acquisition module for acquiring collected measured visible light data; a positioning module for inputting the measured visible light data into a second neural network model trained by the training method of the model for positioning in the first aspect above, and outputting a predicted position corresponding to the measured visible light data.

[0035] In the fifth aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; one or more memories; one or more memories storing one or more programs, which, when one or more programs are executed by one or more processors, enables the electronic device to execute the method of the first or third aspect above.

[0036] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium having instructions stored thereon. When the instructions are executed on a computer, the computer executes the method of the first or third aspect described above.

[0037] In a seventh aspect, an embodiment of the present application provides a computer program product, comprising a computer program / instruction, which implements the method of the first or third aspect mentioned above when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] FIG1 shows an indoor positioning scenario diagram provided by an embodiment of the present application;

[0039] FIG2 is a schematic diagram of a Lambertian model provided in an embodiment of the present application;

[0040] FIG3 is a schematic diagram showing a framework of an attention mechanism provided in an embodiment of the present application;

[0041] FIG4 is a schematic diagram showing the internal structure of an LSTM unit provided in an embodiment of the present application;

[0042] FIG5 a is a schematic diagram showing a process of a fingerprint-based visible light positioning technology;

[0043] FIG5 b and FIG5 c are schematic diagrams showing a process of a visible light positioning technology based on a geometric method;

[0044] FIG5 d is a schematic diagram showing a process of a visible light positioning technology;

[0045] FIG6 a shows a curve diagram of the relationship between RSS and distance provided in an embodiment of the present application;

[0046] FIG6 b is a graph showing the relationship between RSS and distance under artificial light interference provided by an embodiment of the present application;

[0047] FIG6 c is a graph showing the relationship between RSS and distance without artificial light interference provided by an embodiment of the present application;

[0048] FIG6 d is a comparison diagram of the relationship curves between RSS and distance with and without artificial light interference provided by an embodiment of the present application;

[0049] FIG7 is a schematic diagram showing a flow chart of a positioning method provided in an embodiment of the present application;

[0050] FIG8 is a schematic diagram of a data collection scenario in an embodiment of the present application;

[0051] FIG9 is a flow chart showing an offline training process and an online positioning process according to an embodiment of the present application;

[0052] FIG10 is a diagram showing a process of learning / calculating the signal-to-noise ratio and weights at some moments provided in an embodiment of the present application;

[0053] Figures 11a to 11c are schematic diagrams of trajectories collected under different ambient light conditions according to an embodiment of the present application;

[0054] FIG12a and FIG12b are schematic diagrams showing positioning results under interference from artificial light sources provided by an embodiment of the present application;

[0055] FIG13a and FIG13b are schematic diagrams showing comparisons of positioning results under artificial light source interference provided by an embodiment of the present application;

[0056] FIG14a and FIG14b are schematic diagrams showing positioning results under the interference of sunlight provided by an embodiment of the present application;

[0057] FIG15a and FIG15b are schematic diagrams showing comparisons of positioning results under sunlight interference provided by an embodiment of the present application;

[0058] FIG16 is a schematic diagram of a program module of a model training device for positioning provided in an embodiment of the present application;

[0059] FIG17 is a schematic diagram of a program module of a positioning device provided in an embodiment of the present application;

[0060] FIG18 is a schematic diagram showing the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] Before introducing the positioning method provided in the embodiment of the present application, the conceptual terms related to the embodiment of the present application are first introduced.

[0062] Received Signal Strength (RSS) refers to the strength of the signal sent by the transmitter and received by the receiving device. In a visible light positioning system, RSS refers to the visible light signal strength of the LED lights received by the receiving device. It will be understood that in this embodiment of the application, the RSS data of the receiving device 100 is the RSS data of the received signal from each LED light received by the receiving device 100.

[0063] The Lambert model is a light diffuse reflection model that describes an ideal diffuse reflection surface. In the visible light positioning system, it is generally believed that LED lights conform to the Lambert model, as shown in Figure 2. In the Lambert model, it is assumed that the distance d between the receiving device 100 and the LED light is i It is known that the RSS of the signal received by the receiving device 100 is related to multiple factors, such as the radiation angle θ of the signal radiated by the LED lamp and the incident angle ψ of the signal received by the receiving device. The specific relationship can be expressed as the following formula:

[0064]

[0065] Among them, P i (θ, ψ) represents the RSS of the visible light radiated by the i-th LED lamp received by the receiving device 100, m i represents the relevant parameters of the chip of the i-th LED lamp, M represents the relevant parameters of the chip of the receiving device 100, A R Characterizes the effective receiving area of ​​the receiving device 100, P Ti Characterizes the power of the i-th LED lamp, T s (ψ) represents the optical filtering gain of the receiving device 100, and g(ψ) represents the optical concentrator gain of the receiving device 100.

[0066] In the visible light positioning system, since some parameters can be used as known quantities, the above formula (1) can be simplified to obtain a function that represents the relationship between the RSS and the distance between the receiving device 100 and the LED light. The formula is as follows:

[0067] d i =a*RSS b Formula (2);

[0068] Wherein, a and b are Lambertian parameters obtained by simplifying multiple parameters in formula (1).

[0069] The attention mechanism selectively focuses on a portion of the data while processing it, ignoring the rest of the visible data. This allows for the rapid extraction of important features (i.e., attention values) from the data.

[0070] For example, Figure 3 shows a schematic diagram of the framework of an attention mechanism. As shown in Figure 3, Q is a matrix consisting of the queries of the input data in the attention mechanism. Each element in Q represents the query value of each input vector, such as q1, q2, q3, q4 in Figure 3; K is a matrix consisting of the keys of all input vectors in the attention mechanism. Each element in K represents the key value of each input vector, such as k1, k2, k3, k4 in Figure 3; V is a matrix consisting of the values ​​of all input vectors in the attention mechanism. Each element in V represents the value of each input vector, such as v1, v2, v3, v4 in Figure 3. Each key has a corresponding value, forming a data pair for each input vector.

[0071] The calculation process of the attention mechanism is to calculate the correlation between the matrix Q and the matrix K to obtain the weight coefficient of the value corresponding to each key in the matrix K, and then perform weighted summation on multiple values ​​to obtain the final matrix A. Each element in the matrix A represents the attention value (Attention value) of each element in the input data, such as Av1, Av2, Av3, Av4, etc. in Figure 3.

[0072] A Long Short-Term Memory (LSTM) network is a type of recurrent neural network. An LSTM consists of multiple hidden layers (hereinafter referred to as LSTM units), each of which corresponds to data input / output at a specific moment. Hidden layers can transmit both short-term memories (i.e., hidden state, denoted as h) and long-term memories (i.e., cell state, denoted as c), with h also serving as the output of the LSTM unit. In this embodiment, the LSTM can learn from the input RSS data to obtain feature data for each RSS data point.

[0073] In some embodiments, by expanding the LSTM at different times, a schematic diagram of the internal structure of a portion of the LSTM unit can be obtained as shown in FIG4. Taking the LSTM unit 401 as an example, the LSTM unit 401 can receive the c output of the LSTM unit 402 at the previous moment. t-1 、h t-1 And the input x at time t tLSTM unit 401 can use the calculation modules in the unit to convert the c of the previous unit into t-1 Perform selective forgetting and input x into LSTM unit 401 t and h t-1 Perform selective memory and determine the output c of the unit based on the results of selective memory and selective forgetting t and h t .

[0074] For example, as shown in Figure 4, in the selective forgetting process, the input x t and h t-1 Combined with the sigmoid function (i.e., σ in Figure 4), generate the forget gate f t , used to control the c of the previous LSTM unit t-1 Selective forgetting, i.e. t *c t-1 ; In the selective memory process, based on the input x t and h t-1 Combined with the sigmoid function, we can get the selection gate i t , and by selecting gate i t Candidate memory learned based on tanh function in this unit Selective memory, that is During the output process, the result of selective memory can be added to the result of selective forgetting to obtain the output c of the unit. t For the output h t , can be based on the input x t and h t-1 Combined with the sigmoid function to get the output gate o t , and use the output gate to control all the memories learned by the unit (the output c t After conversion by tanh function), the output h of the unit is obtained by filtering t .

[0075] The fully connected (FC) layer performs a weighted sum on the features of the previous layer and maps the features in the feature space to the sample label space through a linear transformation. In this embodiment of the present application, FC can map the features obtained by LSTM from the feature space to the indoor or underground location space to obtain the location of the receiving device 100.

[0076] In visible light positioning systems based on RSS, a light-emitting diode (PD) is typically used as the signal receiving module of the receiving device. The receiving device responds to visible light signals from LED lights, obtains a received signal, modulates the received signal, and then determines the receiving device's position indoors based on the modulation results of multiple LED lights. Existing RSS-based visible light positioning technology is time-consuming and labor-intensive to collect data, and does not consider the impact of ambient light, noise, LED lights, etc. on positioning results, resulting in low positioning accuracy.

[0077] For example, in some embodiments shown in Figure 5a, a fingerprint-based visible light positioning technology is adopted to determine the location of the receiving device 100 by collecting the location fingerprint (such as RSS data) of each sampling point in the scene and the visible light signal actually received by the receiving device 100.

[0078] Taking the visible light positioning system including five LED lights in an indoor scene as an example, the fingerprint-based visible light positioning technology can include two steps: offline database construction and online positioning.

[0079] During the offline database building phase, the map of the indoor space 200 can be divided into N grids, where the grid size can be determined based on the user's positioning accuracy. The location coordinates of each grid and the RSS data of the light data emitted by the LED lights received by the receiving device 100 are collected. A fingerprint database 501 is then built based on each location coordinate and the corresponding RSS data.

[0080] In the online positioning stage, the measured RSS data of the receiving device 100 is matched with multiple sets of RSS data in the fingerprint library 501 through a matching algorithm, and the position of the grid corresponding to the set of RSS data with the highest matching degree is determined as the position of the receiving device 100.

[0081] The matching algorithm may include:

[0082] (1) Euler distance matching method, that is, by calculating the Euler distance between the measured RSS data and the RSS data in the fingerprint library 501, a group of RSS data in the fingerprint library 501 with the smallest Euler distance to the measured RSS data is determined, and the position coordinates corresponding to the group of RSS data are determined to be the actual coordinates of the receiving device 100.

[0083] Among them, the Euler distance can be determined by the following formula:

[0084]

[0085] Among them, i represents the fingerprint number corresponding to different grids in the database, j represents the modulation frequency or number of the LED light, μ i,jis the RSS data received from the LED light with the jth modulation frequency or number by the grid with fingerprint number i in the database, μ o,j It is the RSS data of the reception signal from the LED lamp with the jth modulation frequency or number received by the receiving device 100.

[0086] (2) Deterministic matching method, that is, using the k-nearest neighbor (k-NN) algorithm to determine the k RSS data closest to the measured RSS data, and determine that the position coordinate that corresponds to the same position coordinate the most times among the k nearest RSS data is the measured coordinate.

[0087] (3) Probabilistic matching algorithm: the probability distribution of RSS data is stored in the fingerprint library 501, and then the actual coordinates of the receiving device 100 are calculated using the Bayesian algorithm.

[0088] (4) Nearest neighbor algorithm: This algorithm determines the visible light of the LED light with the largest known coordinates closest to the receiving device 100, and then determines that the actual coordinates of the receiving device 100 match the coordinates of the LED light with the largest RSS data. If the RSS data of multiple LED lights received are the same, the receiving device 100 is determined to be located between the multiple LED lights.

[0089] It is understandable that the fingerprint-based visible light positioning technology requires a large amount of data to establish the fingerprint library 501, and the grid division needs to be very fine to obtain a more accurate positioning result. This technology is time-consuming and labor-intensive.

[0090] In order to solve the problem that the above-mentioned fingerprint-based visible light positioning technology is time-consuming and labor-intensive to implement, in some embodiments shown in Figures 5a and 5b, a visible light positioning technology based on a geometric method is provided, which calculates the distance between the receiving device 100 and multiple LED lights through the Lambertian model to achieve the positioning of the receiving device 100.

[0091] Specifically, as shown in FIG5b , assuming that all LED lights are at the same height and the z coordinate of the receiving device 100 is known, the distances between LED1, LED2, LED3 and the receiving device 100 can be calculated using formula (2) above based on the RSS data of each LED light received by the receiving device 100, completing d1, d2, and d3 respectively. Then, using the three calculated distances as the radius and the corresponding LED light as the center, a circle is drawn, and the point where the three circles intersect is determined to be the position of the receiving device 100. Assuming the position coordinates of the receiving device 100 are (x, y), the coordinates of LED1 are (x1, y1), the coordinates of LED2 are (x2, y2), and the coordinates of LED3 are (x3, y3), the position coordinates of the receiving device 100 can be obtained by solving the following equation:

[0092]

[0093] It is understandable that the circles corresponding to the three LED lights may intersect in an area, such as area S in Figure 5c. In this case, the above equation cannot be used to determine the coordinates of the receiving device 100, or the z coordinate of the receiving device 100 is unknown. In this case, it is necessary to draw a sphere with the LED light as the center. The three spheres may intersect in a three-dimensional space. In this case, to determine the position of the receiving device 100, more visible light emitted by the LED lights is required. To calculate the position coordinates of the receiving device 100, it may be necessary to solve a quadratic or cubic equation of four variables.

[0094] It can be understood that the above-mentioned visible light positioning technology based on geometric methods has a very complicated solution process when calculating the position coordinates of the receiving device 100, and since the intersection points solved based on the distance between the LED light and the receiving device 100 may be multiple or one area, the specific position coordinates cannot be obtained.

[0095] In order to solve the problem that the position coordinate solution process of the visible light positioning technology based on the geometric method is complicated and the specific position coordinates cannot be obtained, a visible light positioning technology based on learning is provided in some embodiments.

[0096] Specifically, similar to the process of fingerprint-based visible light positioning technology, the learning-based visible light positioning technology can include two stages: fingerprint collection and online positioning. However, this technology does not require gridding of indoor maps. The collected fingerprints, such as multiple position coordinates in different motion trajectories and the corresponding RSS data at the position, are used as training data and input into the neural network for error calculation and network parameter optimization.

[0097] Different neural networks can be selected or combined based on different tasks and data characteristics. For example, neural networks can include fully connected layers, convolutional neural networks that can process spatial data, recurrent neural networks that can process time series data, and graph neural networks that can learn topological structures.

[0098] In the online positioning stage, the RSS data collected by the receiving device 100 is input into the trained neural network to obtain the predicted position of the receiving device 100.

[0099] It can be understood that the aforementioned learning-based visible light positioning technology can quickly solve the predicted position coordinates of the receiving device 100 through a neural network. Compared with visible light positioning technology based on geometric methods, the solution process is relatively simple and the solved position coordinates are more specific. However, during the visible light positioning process, the light signal received by the receiving device 100 includes not only the light signal emitted by the LED light, but also ambient light. This technology does not consider the impact of ambient light on visible light positioning, resulting in low positioning accuracy.

[0100] In order to solve the problem of the influence of ambient light on positioning results in the above technology, some embodiments shown in Figure 5d provide a visible light positioning technology, which suppresses the influence of ambient light on positioning results by mapping the measured RSS data into the calibration environment of the Lambertian model.

[0101] Specifically, as shown in FIG5d, by acquiring the RSS data of the receiving device 100 under different ambient light intensities, the relationship between the ambient light intensity and the RSS data of the receiving device 100 is fitted. N (I), the relationship g N (I) can be expressed as the following formula:

[0102]

[0103] Among them, P r,i,N Represents the RSS data of the i-th LED light, I represents the ambient light intensity, I th Indicates the ambient light intensity at the saturation critical point.

[0104] In the positioning stage, the ambient light intensity I collected by the receiving device 100 can be transformed by fast Fourier transform (FFT) to obtain the RSS data P of each LED light. r,i,N , then the ambient light intensity I and RSS data P r,i,N The input parameter adjustment module 511 adjusts the parameters of the Lambert model and applies the adjusted parameters a1 and b1 to the Lambert model 512. Then, based on the RSS data of the receiving signals from each LED lamp received by the receiving device 100, the Lambert model 512 is used to calculate the distance di between the i-th LED lamp and the receiving device 100, thereby realizing the positioning of the receiving device 100.

[0105] It can be understood that the above method also has ambient light in the calibration environment. The ambient light in different calibration environments is different. In the above method, the fitting relationship g N (I) The influence of ambient light in the calibration environment on the Lambertian parameters cannot be eliminated, and the Lambertian model under different calibration ambient light conditions cannot be universal.

[0106] In order to solve the problem that the Lambert model of the above-mentioned visible light positioning technology cannot be universally used under different calibrated ambient lights, an embodiment of the present application provides a model training method for positioning. The model training method for positioning includes: obtaining RSS data and corresponding positions under different ambient lights, as training data, inputting them into a neural network model (i.e., a first neural network model), using the neural network model to learn the impact of different ambient lights on RSS data, learning the relationship between RSS data at the same position, and learning the correspondence between feature data and position, modeling the relationship between RSS data and corresponding position, and obtaining a positioning model (i.e., a second neural network model) that can suppress the influence of ambient light on positioning results and has higher positioning accuracy. Furthermore, the RSS data received by the receiving device of the visible light receiving module can be input into the positioning model as input data, and the positioning model can output the predicted position of the receiving device.

[0107] The positioning model training method provided in the embodiments of the present application trains a neural network model by collecting RSS data under different ambient light conditions. The resulting positioning model can be applied to positioning receiving devices under various complex ambient light interference conditions. Furthermore, because the positioning model is trained using RSS data under different ambient light conditions, it can suppress the influence of ambient light and achieve higher positioning accuracy.

[0108] It is understood that the positioning model in the embodiments of the present application can be applied to positioning scenarios that include multiple visible light sources, including but not limited to environments where satellite signals are weak, such as underground, indoors, and outdoors with many obstructions. Visible light sources include LED lights, energy-saving lamps, incandescent lamps, and other light sources that can emit visible light signals.

[0109] It can be understood that RSS data under different ambient light conditions can be obtained by separating the received data from different LED lights from the received data collected by the receiving device and extracting the corresponding RSS data. The RSS data of multiple visible light sources at the same location under the same ambient light can be used as an RSS data sample. Furthermore, the RSS data sample, the corresponding location, and the corresponding ambient light can be used as a training data sample in the training data, and the training data consisting of multiple training data samples can be input into the neural network model for training.

[0110] In some embodiments, the modulation frequency of the visible light signal of the visible light source may be used as an identifier of the visible light source, and further, FFT may be used to separate light data of different visible light sources.

[0111] It can be understood that the neural network model can learn relevant knowledge of the Lambertian model and locate the position of the visible light source in the scene. The process of learning the neural network model can be understood as a process of optimizing the parameters of the neural network model. The neural network model can include but is not limited to FC layers, convolutional neural network models, recurrent neural network models, graph neural network models, pulse neural network models, etc.

[0112] In some embodiments, the RSS data used as training data may be the RSS data of multiple positions of the receiving device in different motion trajectories. The training data may include RSS data, the time corresponding to the RSS data, the position corresponding to the RSS data, etc. Furthermore, the neural network model may include a neural network model that can process time series data and spatial data, such as a recurrent neural network model (RNN), LSTM, etc. Such a neural network model can not only learn the impact of different ambient light on RSS data, the impact between different LED lights, and the correspondence between RSS data and position, but also learn the relationship between RSS data corresponding to continuous positions on a continuous motion trajectory, for example, the relationship between the continuous changes in RSS data of adjacent positions. The positioning model can also use the learned relationship between the RSS data corresponding to the continuous positions as a constraint condition for the distance between the visible light source and the receiving device. Furthermore, the positioning accuracy of the positioning model obtained through training is higher.

[0113] Furthermore, the neural network model may also include a neural network model layer, such as a FC layer, for mapping feature data output by a neural network model capable of processing both time series and spatial data, such as a recurrent neural network (RNN) or LSTM, to corresponding locations. Specifically, the feature data and corresponding locations may be input into an initial FC layer, which learns the correspondence between the feature data and locations, thereby training the FC layer of the positioning model.

[0114] In some embodiments, the neural network model also includes an attention mechanism. Furthermore, before inputting the training data into the neural network model for training, the attention mechanism can be used to learn the mapping relationship between the RSS data in the training data and the error between the distance observation value obtained based on the Lambertian model and the actual measured distance value, and the attention value of each RSS data is determined based on the mapping relationship.

[0115] Among them, the error is related to the interference of noise, ambient light, etc. during the transmission of the visible light signal. The farther the visible light source, such as an LED lamp, is from the receiving device, the greater the influence of ambient light and noise on the visible light signal emitted by the LED lamp, and the lower the distance accuracy calculated based on the Lambert model. For example, Figure 6a shows the relationship curve between the transmission distance of the visible light signal and the corresponding RSS data, as well as the relationship curve between the transmission distance of multiple noise signals and the corresponding RSS data in some embodiments. The noise includes thermal noise, shot noise and additive white Gaussian noise that interfere with the transmission of the visible light signal. It can be seen from Figure 6a that the RSS of the noise does not change with distance, and the signal-to-noise ratio decreases with increasing distance. The signal-to-noise ratio can reflect the accuracy of the distance calculated by the Lambert model to a certain extent. Combined with the formula (2) of the Lambert model, the distance between the LED lamp and the PD of the receiving device calculated based on the Lambert model can be The error between the actual distance d between the LED lamp and the PD of the receiving device is expressed as the following formula:

[0116]

[0117] Where ε is the channel noise.

[0118] By Taylor expanding the above formula, we can get the following formula:

[0119]

[0120] Here, SNK represents the signal-to-noise ratio. The error is proportional to the true distance d and inversely proportional to the signal-to-noise ratio. Therefore, the attention mechanism can be used to learn the mapping relationship between RSS data and error. The attention value determined based on this mapping relationship can be used as the weight of each RSS data point. After weighting each RSS data point using the attention value, it is input into the neural network model for training. Compared to directly inputting RSS data into the neural network model for training, this can reduce the impact of noise and other factors on positioning results and improve positioning accuracy.

[0121] It can be understood that in some embodiments, in the positioning method provided by the present application, the electronic device for training the positioning model and the receiving device in the online positioning stage can be the same device. In this case, the electronic device includes a visible light receiving module. In other embodiments, in some embodiments, the electronic device and the receiving device can be different devices, wherein the electronic device is used to train the positioning model, and obtain the RSS data received by the receiving device 100 in the online positioning stage, input it into the positioning model, obtain the position of the receiving device 100 and send it to the receiving device. Alternatively, the electronic device can also send the trained positioning model to the receiving device 100, and the receiving device can collect RSS data in the online positioning stage and use the positioning model to determine its own position.

[0122] It can be understood that in some embodiments, for the Lambertian model under different calibration environments, the relationship between the distance between the LED lamp and the receiving device 100 and the corresponding RSS data can be as shown in Figures 6b to 6c.

[0123] FIG6b shows the relationship curve between the distance between the LED lamp and the receiving device 100 and the corresponding RSS data based on the Lambert model under artificial light interference, including the relationship curve obtained by actual measurement (the curve corresponding to the real value in FIG6b ) and the relationship curve calculated by the Lambert model (the curve corresponding to the ideal value in FIG6b ). FIG6c shows the relationship between the distance between the LED lamp and the receiving device 100 and the corresponding RSS data based on the Lambert model without artificial light interference, including the relationship curve obtained by actual measurement (the curve corresponding to the true value in FIG6c ) and the relationship curve calculated by the Lambert model (the curve corresponding to the predicted value in FIG6c ). FIG6b and FIG6c show that the influence of different ambient light on the LED lamp is similar.

[0124] Plotting the true value relationship curves in Figures 6b and 6c in a unified coordinate system yields the graph shown in Figure 6d. It can be seen that the relationship curves corresponding to the two Lambertian models differ in scale. However, by augmenting the relationship curve without artificial light interference (i.e., multiplying the RSS data by a coefficient), the two relationship curves can essentially overlap. Therefore, by augmenting RSS data collected under a specific calibration environment, RSS data under other ambient lighting conditions can be obtained.

[0125] Therefore, in some embodiments, RSS data under a certain ambient light environment can be obtained, and RSS data under different ambient lights can be constructed through data enhancement. The enhanced RSS data, the RSS data under the calibration environment, and the corresponding positions are used as training data.

[0126] It is understood that in some embodiments, during the training process of the positioning model, a loss function of the positioning model can be constructed based on the predicted position obtained by the trained positioning model and the corresponding true position, and the error back propagation is performed based on the loss function to optimize the parameters of the positioning model. Furthermore, the instability caused by random errors can be minimized in the traditional loss function, for example, by adding a regularization term related to the distance error to the loss function. It is understandable that the traditional loss function may cause overfitting of the positioning model. Back propagation is performed based on the updated loss function to optimize the parameters of the positioning model, which can enhance the anti-interference ability (i.e., robustness) of the positioning model.

[0127] 7 , the model training method for positioning in the embodiment of the present application is described in detail by taking the visible light source as an LED lamp and the receiving device receiving the visible light signal through a PD as an example.

[0128] It's understood that before executing the following method, you need to first establish an RSS data collection environment. Specifically, within a specific location, such as indoor space 200 in Figure 1, install multiple LED lights and calibrate their positions and heights, as well as the height of the PD. For ease of description, the following example uses the installation of five LED lights. You also need to set LED light identifiers, such as their modulation frequencies. This allows you to separate the data from different LED lights within the PD's received data and extract the RSS data corresponding to each LED light.

[0129] FIG7 is a flow chart illustrating a positioning method according to an embodiment of the present application. It is understood that the execution subject in FIG7 is an electronic device. It is understood that the present application uses the example of a PD receiving visible light located in an electronic device. In other embodiments, the PD may be located in another visible light receiving device outside the electronic device.

[0130] As shown in FIG7 , the method includes:

[0131] 701: Obtain RSS data of multiple locations measured under different ambient light conditions, and use the obtained RSS data, the corresponding locations, and the corresponding ambient light conditions as training data.

[0132] It can be understood that the RSS data measured under different ambient light conditions can be obtained by separating the received data from different LED lights from the received data collected by the receiving device and extracting the corresponding RSS data. The RSS data of multiple visible light sources at the same location under the same ambient light can be used as an RSS data sample. Furthermore, the RSS data sample, the corresponding location, and the corresponding ambient light can be used as a training data sample in the training data, and the training data consisting of multiple training data samples can be input into the neural network model for training.

[0133] In some embodiments, the modulation frequency of the visible light signal of the LED lamp can be used as the identification of the LED lamp, and then, FFT can be used to separate the received data of different LED lamps.

[0134] In some embodiments, RSS data under different ambient light conditions can be obtained by performing data enhancement on RSS data under a certain ambient light condition.

[0135] For example, the RSS data of the i-th position obtained under a certain ambient light can be expressed as: Wherein, N represents the number of LED lights in the test environment. In this embodiment, N is 5. j represents the position number corresponding to the RSS data. (x j ,y j ) represents the location coordinates corresponding to the RSS data, Indicates the RSS data of LED light numbered i received at position numbered j.

[0136] Furthermore, data enhancement of the above RSS data may include: generating a random parameter λ∈(0,α], where α is a preset value and can be adjusted according to actual conditions. By performing data enhancement on the above RSS data through random parameters, RSS data under other ambient light conditions can be obtained, for example Multiple λ values ​​can be randomly generated to obtain RSS data under multiple ambient light conditions.

[0137] 702: Initialize the neural network model.

[0138] It is understood that initializing the neural network model can be to set the initial parameters of each key node in the neural network model. The neural network model can learn relevant knowledge of the Lambertian model and the position of the LED light in the positioning scene.

[0139] It can be understood that the neural network model may include but is not limited to FC layer, convolutional neural network model, recurrent neural network model, graph neural network model, pulse neural network model, etc.

[0140] In some embodiments, the RSS data used as training data may be the RSS data of multiple positions of the PD in different motion trajectories, and each training data sample in the training data may include an RSS data sample, the time corresponding to the RSS data sample, the position corresponding to the RSS data sample, etc. Furthermore, the neural network model in step 702 may include a neural network model capable of processing time series data and spatial data, such as a recurrent neural network model (RNN), LSTM, etc. Such a neural network model can not only learn the spatial characteristics of the training data, such as the impact of different ambient lights on the RSS data, the relationship between the RSS data at the same position (i.e., the correlation between different LED lights), and the correspondence between feature data and position, but can also learn the spatial characteristics of the training data related to time, such as the relationship between RSS data corresponding to consecutive positions on a continuous motion trajectory, etc.

[0141] As you can understand, since the RSS data of adjacent locations changes continuously, the positioning model can also use the learned relationship between the RSS data corresponding to consecutive locations as a constraint on the distance between the LED light and the PD. As a result, the trained positioning model has higher positioning accuracy.

[0142] For example, FIG8 is a schematic diagram of a data collection scenario in an embodiment of the present application.

[0143] As shown in FIG8 , five LED lights are pre-installed in the indoor space 801 , and the electronic device including the PD can be placed on the movable robot 802 to collect the received data collected at multiple position points when the movable robot moves along the trajectory 803 , where each circle can correspond to one position point.

[0144] It can be understood that, taking multiple points in trajectory segment 804 as an example, points 8041, 8042, and 8043 are three consecutive points on trajectory segment 804. The received data collected at points 8041, 8042, and 8043 changes continuously and has a certain correlation. The RSS data corresponding to points 8041, 8042, and 8043, as well as the corresponding positions and times of the RSS data, are input into the neural network model for training. The neural network model can then learn the changing relationship of this RSS data. When performing position prediction, this learned changing relationship can be used as a constraint condition for the distance between the LED light and the PD.

[0145] Furthermore, the neural network model also includes a neural network model layer, such as an FC layer, that can be used to map feature data output by the aforementioned neural network models capable of processing time series data and spatial data, such as recurrent neural network models (RNNs) and LSTMs, to corresponding locations. Specifically, the feature data and corresponding locations can be input into an initial FC layer, which can learn the correspondence between the feature data and locations, thereby training the FC layer of the positioning model.

[0146] In some embodiments, the neural network model also includes an attention mechanism. It can be understood that the error between the distance observation value obtained by the Lambert model and the measured distance value is related to the interference of noise, ambient light, etc. during the transmission of the visible light signal. The farther the visible light source, such as an LED lamp, is from the receiving device, the greater the influence of ambient light and noise on the visible light signal emitted by the LED lamp, and the lower the distance accuracy calculated based on the Lambert model. Before the training data is input into the LSTM layer for training, the attention mechanism can be used to learn the mapping relationship between the RSS data in the training data and the error between the distance observation value obtained based on the Lambert model and the measured distance value, and determine the attention value of each RSS data based on the mapping relationship. The attention value determined based on the mapping relationship can be used as the weight of each RSS data. Furthermore, each RSS data can be weighted by the attention value and input into the LSTM layer for training, which can reduce the influence of noise and the like on the output of the feature data of the LSTM layer and obtain feature data of RSS data with smaller error.

[0147] Furthermore, in some embodiments, the attention mechanism can be a multilayer perceptron (MLP) with an attention mechanism. The MLP may include an input layer, an output layer, and a hidden layer connecting the input layer and the output layer. The input layer is used to obtain the input RSS data, and the hidden layer is used to learn the correlation between the input RSS data and the error between the distance observation value and the measured distance value, obtain the attention value of the RSS data, and output the attention value through the output layer. The attention value output by the output layer can be converted to between (0, 1) through a softmax function to obtain the weight of each RSS data.

[0148] 703: Input the training data into the neural network model for training to obtain a positioning model.

[0149] It is understood that in some embodiments, the neural network model may include an LSTM layer and an FC layer. Inputting the training data into the neural network model may include: dividing the RSS data samples corresponding to the same ambient light and motion trajectory in the training data into a training data subset, and inputting the RSS data in each training data subset into the LSTM layer in chronological order for training.

[0150] For example, in the LSTM units corresponding to each moment in the LSTM layer shown in Figure 4, for RSS data samples in the same training data subset, the LSTM units can learn the relationships between RSS data samples at the same moment in different training data subsets, as well as the characteristic relationships between RSS data samples at multiple moments. For RSS data in different training data subsets, the LSTM units can reinforce and optimize the learned relationships between RSS data samples at the same moment and the characteristic relationships between RSS data samples at multiple moments. Furthermore, the LSTM units can learn the relationships between RSS data under different ambient light conditions, thereby training an LSTM layer that is independent of ambient light and capable of extracting RSS data features. The feature data output by the LSTM layer and the corresponding position of each feature data can serve as input data for the FC layer, which learns the relationship between feature data and position, resulting in an FC layer capable of mapping RSS data features to positions in the positioning scenario.

[0151] In some embodiments, the neural network model may further include an attention mechanism. Furthermore, before the training data is input into the LSTM layer for training, the attention mechanism may be used to learn the mapping relationship between the RSS data in the training data and the error between the distance observation value obtained based on the Lambertian model and the actual distance value, and determine the attention value of each RSS data based on this mapping relationship.

[0152] Furthermore, in some embodiments, the attention mechanism can be an MLP with an attention mechanism. The MLP can obtain RSS data samples and learn the correlation between the RSS data samples and the errors between the distance observation values ​​and the measured distance values ​​obtained by the Lambert model, thereby obtaining an MLP that can calculate the attention value of each RSS data in the RSS data sample. Furthermore, the LSTM layer can convert the attention value output by the MLP through a softmax function to obtain the weight of each RSS data. In addition, the weight of the RSS data can be used to weight the corresponding RSS data to reduce the influence of noise and other interferences on the training of the LSTM layer. That is, before the LSTM is trained, data preprocessing can be performed through MLP, softmax function, etc., and the weighted RSS data can be used as training data for LSTM training.

[0153] In some embodiments, the above-mentioned conversion of the attention value output by the MLP through the softmax function to obtain the weight of each RSS data, and the use of the weight of the RSS data to weight the corresponding RSS data can be implemented through other data preprocessing modules instead of the LSTM layer for data processing, and this application does not impose any restrictions on this.

[0154] 704: Obtain the RSS data to be located.

[0155] It can be understood that the RSS data to be located is the received data received by the PD of the electronic device during the online positioning stage, and the RSS data corresponding to each LED light in the positioning space is obtained by separating and processing the received data based on the identification of different LED lights.

[0156] In some embodiments, the modulation frequency of the LED lamp is used as the identification of each LED lamp. After the received data is obtained, the received data can be separated by FFT to obtain the RSS data corresponding to each LED lamp.

[0157] 705: Input the RSS data to be located into the positioning model, and output the predicted position of the RSS data to be located.

[0158] It can be understood that the positioning model can use the attention mechanism to calculate the attention value of the RSS data corresponding to different LED lights in the RSS data to be located, and convert the attention value into the weight of each RSS data through the softmax function. The positioning model can input the weighted RSS data into the trained LSTM layer, calculate the feature data of the weighted RSS data, and map the RSS data to a certain position in the positioning space through the FC layer to obtain the predicted position of the RSS data to be located.

[0159] It can be understood that the above steps 701 to 703 are the offline training process in the positioning method provided in the embodiment of the present application, and steps 704 and 705 are the online positioning process in the positioning method provided in the embodiment of the present application. In some embodiments, the embodiment of the present application may be a model training method for positioning including the above steps 701 to 703. In other embodiments, the embodiment of the present application may be a positioning method including the above steps 704 to 705. In other embodiments, the embodiment of the present application may be a positioning method including the above steps 701 to 705. This application is not limited to this.

[0160] The positioning method provided in the embodiments of the present application collects RSS data under different ambient light conditions to train a neural network model. The resulting positioning model can be applied to positioning receiving devices under various complex ambient light interference conditions. Furthermore, because the positioning model is trained using RSS data under different ambient light conditions, it can suppress the influence of ambient light and achieve higher positioning accuracy.

[0161] In some embodiments, during the training of the positioning model in step 703, a portion of the RSS data in the training data and the corresponding true location can be used as optimization data to optimize the trained positioning model. Specifically, the optimization data can be input into the trained positioning model, and the predicted location and the corresponding true location can be used to construct a loss function for the positioning model. Error backpropagation is then performed based on the loss function to optimize the parameters of the positioning model.

[0162] In some embodiments, the constructed loss function can be expressed by the following formula:

[0163]

[0164] Among them, p i Indicates the true position of PD, represents the i-th predicted position determined by the positioning model.

[0165] Furthermore, the instability caused by random errors can be minimized in traditional loss functions, for example, by adding a regularization term related to distance error. It is understandable that traditional loss functions may cause the positioning model to overfit, affecting its stability. Backpropagation based on the updated loss function and optimizing the positioning model parameters can enhance the positioning model's anti-interference ability (i.e., robustness).

[0166] In other embodiments, the constructed wheel function with the regularization term added can be expressed by the following formula:

[0167]

[0168] Among them, β is the regularization adjustment coefficient of the regularization term, N is the number of LED lights, d j Indicates the distance between the PD and the jth LED lamp calculated based on the Lambert model, RSS j is the RSS data corresponding to the j-th LED light received by the PD, ε represents the channel noise, and ε~N(μ,δ) is Gaussian noise. The mean and variance of Gaussian noise can be estimated from the RSS data.

[0169] The offline training process and the online positioning process in FIG. 7 are further described below in conjunction with FIG. 9 .

[0170] Figure 9 is a flow chart of an offline training process and an online positioning process provided by an embodiment of the present application. It is understood that the dotted arrows indicate the data flow direction of the offline training process, the thin solid arrows indicate the data flow direction of the online positioning process, and the thick solid arrows indicate the data flow direction common to both the offline training process and the online positioning process.

[0171] It can be understood that in the offline training process, the weight learning module 930, the interference suppression module 940 and the error calculation module 950 constitute a neural network model; in the online positioning process, the weight learning module 930, the interference suppression module 940 (and the error calculation module 950) constitute a positioning model.

[0172] As shown in Figure 9, the offline training process includes:

[0173] The signal receiving module (offline) 910 obtains a PD reception value 911 (i.e., the PD reception data mentioned above) through the PD. The signal receiving module (offline) 910 can separate the reception data corresponding to different LED lights in the PD reception value through FFT to obtain the RSS data 912 of each LED light. The signal receiving module (offline) 910 can also expand the RSS data under a single ambient light through data enhancement to obtain expanded RSS data 913. The signal receiving module (offline) 910 can input the RSS data 912 and the expanded RSS data 913 as training data into the signal-to-noise ratio (SNR) calculation unit 931 of the weight learning module 930 and the noise characteristic learning unit 941 of the interference suppression module, respectively.

[0174] The SNK calculation unit 931 calculates the signal-to-noise ratio (SNR) of the training data and inputs the calculated SNR into the attention layer 932 along with the training data as a query. This layer learns the mapping between RSS data and observed distance error. The attention layer 932 then outputs an attention value for each RSS data point in the training data. The observed distance error is the error between the observed distance between the LED light and the PD calculated using the Lambertian model and the actual distance. The output attention value from the attention layer 932 is then input into the noise characteristic learning unit 941.

[0175] The noise characteristic learning unit 941 can convert the attention value of each RSS data into a weight for each RSS data. For example, the attention value can be converted to a value between (0, 1) by using a softmax function. Furthermore, the noise characteristic learning unit 941 can use the calculated weight to weight the RSS data in the training data and input the weighted RSS data into the LSTM layer 942.

[0176] LSTM layer 942 can learn the relationship between RSS data at the same time in the weighted training data, the characteristic relationship between RSS data samples at multiple times, the relationship between RSS data under different ambient light conditions, and the changing relationship between RSS data at different locations on the same motion trajectory, thereby optimizing the parameters of LSTM layer 942. LSTM layer 942 can output RSS data and characteristic data of the corresponding location, and input the characteristic data into FC layer 943.

[0177] The FC layer 943 may learn the mapping relationship between the feature data of the RSS data and the feature data of the corresponding position, and optimize the parameters of the FC layer.

[0178] In some embodiments, the noise characteristic learning unit 941, LSTM layer 942 and FC layer 943 of the interference suppression module, and the parameters of the attention layer 932 of the weight learning module 930 can be optimized based on the back propagation of the loss error calculated by the error calculation module 930.

[0179] Specifically, the error calculation module 930 can construct a loss function 953 based on the position 951 predicted by the positioning model output by the FC layer and its corresponding true position 952. The loss function can be, for example, the loss function corresponding to formula (7) or formula (8). For the constructed loss function 953, the error calculation module 930 can minimize the loss function to obtain a loss error 954, and then reversely propagate the loss error layer by layer to optimize the aforementioned neural network model layer to improve positioning accuracy.

[0180] As shown in Figure 9, the online positioning process includes:

[0181] The signal receiving module (online) 920 obtains a PD reception value 921 (i.e., the PD reception data described above) through the PD. The signal receiving module (online) 920 can separate the reception data corresponding to different LED lights in the PD reception value through FFT to obtain RSS data 922 for each LED light. The signal receiving module (online) 920 can input RSS data 922 into the signal-to-noise ratio (SNR) calculation unit 931 of the weight learning module 930 and the noise characteristic learning unit 941 of the interference suppression module.

[0182] The SNK calculation unit 931 can calculate the signal-to-noise ratio of the RSS data 922, and input the calculated signal-to-noise ratio into the attention layer 932 together with the RSS data 922 as a query of the attention layer 932, and output an attention value for each RSS data in the RSS data 922. The attention layer 932 can input the output attention value into the noise characteristic learning unit 941.

[0183] The noise characteristic learning unit 941 can convert the attention value of the RSS data into a weight for the RSS data 922. For example, the attention value can be converted to a value between (0, 1) using a softmax function. Furthermore, the noise characteristic learning unit 941 can weight the RSS data 922 using the calculated weight and input the weighted RSS data into the LSTM layer 942.

[0184] The LSTM layer 942 may perform calculations based on the input weighted RSS data, output feature data of the RSS data, and input the feature data into the FC layer 943 .

[0185] The FC layer 943 may map the feature data of the RSS data to the position space in the positioning scenario to obtain the predicted position corresponding to the RSS data 922 .

[0186] In some embodiments, the attention layer may be an MLP, and the PD reception value may be reception data of multiple position points collected along the PD motion trajectory. The PD reception value may then be input into the weight learning module 930 and the interference suppression module 940 in chronological order to perform signal-to-noise ratio and weight learning / calculation. For example, FIG10 shows a process diagram of signal-to-noise ratio and weight learning / calculation at some moments provided in an embodiment of the present application.

[0187] Taking the offline training process as an example, for the PD received value, the RSS data corresponding to each LED light can be obtained through FFT transformation, and the RSS data after FFT is respectively input into the SNK calculation unit 931 and the input noise characteristic learning unit 941. The SNK calculation unit 931 can obtain the signal-to-noise ratio of the RSS data through the spectrum diagram after FFT, and input the SNR and RSS data into the MLP to learn the mapping relationship between the SNR and RSS data to obtain the attention value of the RSS data.

[0188] The noise characteristic learning unit 941 may convert the attention value of the received RSS data into the weight of the RSS data, weight the RSS data, and input the weighted RSS data into the LSTM layer.

[0189] It can be understood that the positioning model trained by the offline training method in Figure 9 can obtain a positioning model that is more trustworthy in the RSS data of the LED lights closest to the PD through the weight learning module 930 and the noise characteristic learning unit 941, which can effectively suppress noise interference and reduce the impact of the RSS data corresponding to the interfered LED lights on the predicted position. In addition, by learning the feature data of the RSS data under different ambient light through the LSTM layer 942, a positioning model that can effectively suppress ambient light interference can be obtained, which can achieve high-precision positioning under complex ambient light. In addition, by adding an error regularization term to the error function constructed by the error calculation module 950, the robustness of the positioning model in the face of Gaussian noise can be enhanced, and the stability of the positioning model can be higher.

[0190] In order to better illustrate the beneficial effects of the positioning model training method provided in the embodiment of the present application, the positioning environment of the positioning scenario in Figure 1 is taken as an example, combined with a specific test environment and test data for explanation.

[0191] It is understood that in some embodiments, the training data collection environment is built to obtain a positioning environment such as the positioning scene in Figure 1. Assume that the positioning space in Figure 1 is an empty room with a size of 5*5*2.6m 3 The visible light signal transmitter, consisting of an LED lamp, an STM32 processor, and a pulse width modulation (PWM) module, was installed at the top of the positioning space shown in Figure 1. Five transmitters were installed. The LED lamps served as the visible light source, with a power of 10W. The visible light signals emitted by the LED lamps of the transmitters were modulated at 1756.1kHz, 2571kHz, 3137.3kHz, 4235.3kHz, and 4864.9kHz, respectively. The coordinates of the transmitters were (0.26, 0.66, 2.6), (4.61, 0.67, 2.6), (4.57, 4.25, 2.6), (0.18, 4.25, 2.6), and (2.45, 2.49, 2.6).

[0192] It can be understood that the settings of various parameters in the test environment are shown in Table 1 below:

[0193] Table 1

[0194]

[0195] In the aforementioned calibration test environment, trajectory data was collected for conditions with sunlight interference, artificial light interference, and visible light signals from LEDs alone. Two RPD received values ​​were collected for both conditions, as shown in Figures 11a and 11b. Three PD received values ​​were collected for conditions with visible light signals from LEDs alone, as shown in Figure 11c. This collected data served as training data, and the training method described in steps 701 to 703 in Figure 7 was used to generate a positioning model.

[0196] In order to illustrate the beneficial effects of the model training method for positioning provided in the embodiment of the present application, some real trajectories are also collected in the embodiment of the present application, and positioning is performed using the positioning model to obtain a predicted trajectory composed of the predicted position output based on the RSS data collected from the real trajectory. The details are as follows:

[0197] As shown in Figure 12a, there are the actual trajectory of trajectory 1 collected under the interference of artificial light source (i.e., the trajectory corresponding to the true value in Figure 12a), the positioning trajectory corresponding to the RSS data of trajectory 1 output by the positioning model provided in an embodiment of the present application (i.e., the trajectory corresponding to the positioning model in Figure 12a), and the predicted trajectory corresponding to the RSS data of trajectory 1 output by the traditional visible light positioning method (such as the geometry-based visible light positioning method, etc.) (the trajectory corresponding to the traditional positioning in Figure 12a).

[0198] As shown in Figure 12b, there are the actual trajectory of trajectory 2 collected under the interference of artificial light source (i.e., the trajectory corresponding to the true value in Figure 12b), the positioning trajectory corresponding to the RSS data of trajectory 2 output by the positioning model provided in the embodiment of the present application (i.e., the trajectory corresponding to the positioning model in Figure 12b), and the predicted trajectory corresponding to the RSS data of trajectory 2 output by the traditional positioning method (the trajectory corresponding to the traditional positioning in Figure 12b).

[0199] Based on the relationship between the ranging error and the cumulative error (CDF) between the positioning trajectory output by the positioning model in the embodiment of the present application and the actual trajectory, as well as the relationship between the predicted trajectory output by the traditional positioning method and the actual trajectory, the relationship comparison curve of trajectory 1 in Figure 13a and the relationship comparison curve of Figure 13b can be drawn respectively.

[0200] Combining the relevant data of Figures 12a to 13b, the following Table 2 can be obtained:

[0201] Table 2

[0202]

[0203] Wherein, ME is the mean error (ME), RMSE is the root mean square error (RMSE), and 50% and 90% respectively indicate that the positioning errors are 50% and 90%.

[0204] As shown in Figure 14a, there are the true trajectory of trajectory 1 collected under the interference of sunlight (i.e., the trajectory corresponding to the true value in Figure 14a), the predicted trajectory corresponding to the RSS data of trajectory 1 output by the positioning model provided in an embodiment of the present application (i.e., the trajectory corresponding to the positioning model in Figure 14a), and the trajectory diagram corresponding to the RSS data of trajectory 1 output by the traditional positioning method (i.e., the trajectory corresponding to the traditional positioning in Figure 14a).

[0205] As shown in Figure 14b, this is a trajectory diagram of the actual trajectory of trajectory 2 collected under interference from sunlight (i.e., the trajectory corresponding to the true value in Figure 14b), the predicted trajectory corresponding to the RSS data of trajectory 2 output by the positioning model provided in an embodiment of the present application (i.e., the trajectory corresponding to the positioning model in Figure 14b), and the trajectory corresponding to the RSS data of trajectory 2 output by the traditional positioning method (i.e., the trajectory corresponding to the traditional positioning in Figure 14b).

[0206] The relationship between the ranging error and the cumulative error (CDF) between the positioning trajectory output by the positioning model in the embodiment of the present application and the actual trajectory, as well as the relationship between the predicted trajectory output by the traditional positioning method and the actual trajectory, can be plotted as the relationship comparison curve of trajectory 1 in Figure 15a and the relationship comparison curve in Figure 15b respectively.

[0207] Combining the relevant data of Figures 14a to 15b, the following Table 3 can be obtained:

[0208] Table 3

[0209]

[0210] Wherein, ME is the mean error (ME), RMSE is the root mean square error (RMSE), and 50% and 90% respectively indicate that the positioning errors are 50% and 90%.

[0211] Taking the data corresponding to trajectory 1 in Figures 12a and 13a and Table 2 as an example, Table 2 shows that the 50% and 90% positioning errors of the positioning model are 10.9 cm and 23.55 cm, respectively. Compared to the 50% and 90% positioning errors of 38.8 cm and 55 cm of the traditional positioning method, these are reductions of 71.39% and 57.22%, respectively. Furthermore, the average positioning error of the positioning model is 12.48 cm, a 68.60% reduction compared to the 39.07 cm of the traditional positioning method. Similar improvements are also seen in data corresponding to other trajectories under ambient light interference.

[0212] It can be understood that the positioning model trained by the positioning model training method provided in the embodiment of the present application can effectively suppress the interference of ambient light and noise compared to the traditional positioning method, and has higher positioning accuracy.

[0213] To solve the problem that the Lambert model of the visible light positioning technology mentioned above cannot be universally applied under different calibration ambient light conditions, an embodiment of the present application provides a model training device for positioning. As shown in FIG16 , the model training device 1600 for positioning includes: a data acquisition module 1601 and a model training module 1602 .

[0214] The data acquisition module 1601 is configured to acquire a plurality of visible light data collected at a plurality of locations under different ambient light conditions, wherein each visible light data includes light data from a different visible light source.

[0215] Among them, the model training module 1602 is used to input each visible light data, the position information corresponding to each visible light data, and the ambient light identification of the ambient light corresponding to each visible light data as training data samples into the first neural network model to train the first neural network model and obtain the second neural network model.

[0216] It can be understood that in some embodiments, the above-mentioned data acquisition module 1601 is equivalent to the signal receiving module (offline) 910 of the offline training process in Figure 9, and the model training module 1602 is equivalent to the weight learning module 930, interference suppression module 940 and error calculation module 950 of the offline training process in Figure 9.

[0217] To solve the problem that the above-mentioned visible light positioning technology cannot be universally applied to the Lambertian model under different calibration ambient light conditions, an embodiment of the present application provides a positioning device. As shown in FIG17 , the positioning device 1700 includes: a data acquisition module 1701 and a positioning module 1702 .

[0218] The data acquisition module 1701 is used to obtain the collected measured visible light data.

[0219] Among them, the positioning module 1702 is used to input the measured visible light data into the second neural network model trained by the model training device 1600 for positioning in Figure 16 above, and output the predicted position corresponding to the measured visible light data.

[0220] The hardware structure of the electronic device in the embodiment of the present application is described below with reference to FIG18 .

[0221] FIG18 shows a hardware block diagram of an electronic device 100 that applies the method provided in an embodiment of the present application. The electronic device 100 can be used to run the instructions related to the positioning method provided in the embodiment of the present application. In FIG18 , similar components have the same reference numerals. As shown in FIG18 , the electronic device 100 may include a processor 110, a power module 140, a memory 180, a camera 170, a mobile communication module 130, a wireless communication module 120, a sensor module 190, an audio module 150, an interface module 160, a button 101, and a display screen 102, etc. In an embodiment of the present application, the memory 180 can be used to store the instructions related to the page display method provided in the embodiment of the present application and the intermediate data files generated when the processor executes the instructions.

[0222] It should be understood that the structure illustrated in the embodiments of the present invention does not constitute a specific limitation on the electronic device 100. In other embodiments of the present application, the electronic device 100 may include more or fewer components than shown, or may combine or separate certain components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0223] It is understood that the electronic device in the embodiment of the present application can be a mobile phone including a PD, a portable android device (PAD), a personal digital assistant (PDA), a handheld device with a wireless communication function, a computing device, a vehicle-mounted device or a wearable device, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a wireless terminal in unmanned driving, a wireless terminal in remote medical care, a wireless terminal in a smart grid, a wireless terminal in transportation safety, a wireless terminal in a smart city, a wireless terminal in a smart home, and the like, and can also be an on-chip integrated chip including a PD and a processor. The form of the electronic device in the embodiment of the present application is not specifically limited.

[0224] The processor 110 may include one or more processing units, for example, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor DSP, a microprocessor (MCU), an artificial intelligence (AI) processor or a programmable logic device (Field Programmable Gate Array, FPGA) and the like. The different processing units may be independent devices or integrated into one or more processors. A storage unit may be provided in the processor 110 for storing instructions and data. In some embodiments, the storage unit in the processor 110 is a cache memory 180. The processor may be used to execute the relevant method instructions in Figure 7 or Figure 9 provided in the embodiments of the present application.

[0225] The power module 140 may include a power supply, a power management component, and the like. The power supply may be a battery. The power management component manages the charging of the power supply and the supply of power to other modules. In some embodiments, the power management component includes a charging management module and a power management module. The charging management module receives charging input from a charger; the power management module connects the power supply, the charging management module, and the processor 110. The power management module receives input from the power supply and / or the charging management module to power the processor 110, the display 102, the camera 170, and the wireless communication module 120.

[0226] The mobile communication module 130 may include but is not limited to an antenna, a power amplifier, a filter, an LNA (Low noise amplify, low noise amplifier), etc. The mobile communication module 130 can provide solutions for wireless communications including 2G / 3G / 4G / 5G applied to the electronic device 100. The mobile communication module 130 can receive electromagnetic waves through the antenna, filter, amplify and other processing on the received electromagnetic waves, and transmit them to the modulation and demodulation processor for demodulation. The mobile communication module 130 can also amplify the signal modulated by the modulation and demodulation processor, and convert it into electromagnetic waves for radiation through the antenna. In some embodiments, at least some functional modules of the mobile communication module 130 can be set in the processor 110. In some embodiments, at least some functional modules of the mobile communication module 130 can be set in the same device as at least some modules of the processor 110. Wireless communication technologies may include global system for mobile communications (GSM), general packet radio service (GPRS), code division multiple access (CDMA), wideband code division multiple access (WCDMA), time-division code division multiple access (TD-SCDMA), long term evolution (LTE), Bluetooth (BT), global navigation satellite system (GNSS), wireless local area networks (WLAN), near field communication (NFC), frequency modulation (FM) and / or field communication (NFC), infrared technology (IR), etc.GNSS may include the global positioning system (GPS), the global navigation satellite system (GLONASS), the Beidou navigation satellite system (BDS), the quasi-zenith satellite system (QZSS) and / or the satellite based augmentation system (SBAS).

[0227] The wireless communication module 120 may include an antenna and realize the transmission and reception of electromagnetic waves via the antenna. The wireless communication module 120 may provide wireless communication solutions including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication technology (NFC), infrared technology (IR), etc., which are applied to the electronic device 100. The electronic device 100 can communicate with the network and other devices through wireless communication technology.

[0228] In some embodiments, the mobile communication module 130 and the wireless communication module 120 of the electronic device 100 may also be located in the same module.

[0229] Display screen 102 is used to display human-computer interaction interfaces, images, videos, etc. Display screen 102 includes a display panel. The display panel can be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode (AMOLED), a flexible light-emitting diode (FLED), a MiniLED, a MicroLED, a Micro-oLed, or a quantum dot light-emitting diode (QLED).

[0230] The sensor module 190 may include a proximity sensor, a pressure sensor, a gyro sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a distance sensor, a fingerprint sensor, a temperature sensor, a touch sensor, an ambient light sensor, a bone conduction sensor, and the like.

[0231] The audio module 150 is used to convert digital audio information into analog audio signal output, or convert analog audio input into digital audio signal. The audio module 150 can also be used to encode and decode audio signals. In some embodiments, the audio module 150 can be provided in the processor 110, or some functional modules of the audio module 150 can be provided in the processor 110. In some embodiments, the audio module 150 can include a speaker, an earpiece, a microphone, and a headphone jack.

[0232] Camera 170 is used to capture still images or videos. The lens generates an optical image of an object and projects it onto a photosensitive element. The photosensitive element converts the optical signal into an electrical signal, which is then passed to the image signal processing (ISP) for conversion into a digital image signal. Electronic device 100 implements a camera function using the ISP, camera 170, video codec, GPU (Graphics Processing Unit), display 102, and application processor.

[0233] The interface module 160 includes an external memory interface, a universal serial bus (USB) interface, and a subscriber identification module (SIM) card interface. The external memory interface can be used to connect an external memory card, such as a Micro SD card, to expand the storage capacity of the electronic device 100. The external memory card communicates with the processor 110 via the external memory interface to implement data storage. The USB interface is used for the electronic device 100 to communicate with other electronic devices. The subscriber identification module card interface is used to communicate with a SIM card installed in the electronic device 10010, for example, to read a phone number stored in the SIM card or to write a phone number to the SIM card.

[0234] In some embodiments, the electronic device 100 further includes a button 101, a motor, and an indicator. The button 101 may include a volume button, an on / off button, and the like. The motor is used to vibrate the electronic device 100, for example, when the user's electronic device 100 is called, to prompt the user to answer the call. The indicator may include a laser pointer, a radio frequency indicator, an LED indicator, and the like.

[0235] References in the specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one exemplary implementation or technique disclosed herein. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0236] The present disclosure also relates to a device for performing the operations in the text. The device can be constructed specifically for the required purpose or it can include a general-purpose computer that is selectively activated or reconfigured by a computer program stored in the computer. Such a computer program can be stored in a computer-readable medium, such as, but not limited to, any type of disk, including a floppy disk, an optical disk, a CD-ROM, a magneto-optical disk, a read-only memory (ROM), a random access memory (RAM), an EPROM, an EEPROM, a magnetic or optical card, an application-specific integrated circuit (ASIC), or any type of medium suitable for storing electronic instructions, and each can be coupled to a computer system bus. In addition, the computer mentioned in the specification can include a single processor or can be an architecture involving multiple processors for increased computing power.

[0237] Additionally, the language used in this specification has been primarily selected for readability and instructional purposes and may not have been selected to delineate or circumscribe the disclosed subject matter.Thus, this disclosure is intended to illustrate but not to limit the scope of the concepts discussed herein.

Claims

1. A training method for a positioning model, characterized in that: include: Acquire a plurality of visible light data collected at a plurality of locations under different ambient light conditions at each location, wherein each piece of visible light data includes light data from a different visible light source; Each visible light data, the position information corresponding to each visible light data, and the ambient light identification of the ambient light corresponding to each visible light data are used as training data samples and input into the first neural network model to train the first neural network model and obtain a second neural network model.

2. The method for training a positioning model according to claim 1, wherein: The position information corresponding to the visible light data is the coordinates of the position of the visible light data obtained by taking a preset position in the positioning scene as the origin of a spatial coordinate system.

3. The method for training a positioning model according to claim 1, wherein: The acquiring of a plurality of visible light data collected at a plurality of positions under different ambient light conditions at each position includes: Obtaining visible light data at each position collected at multiple positions under preset ambient light; Data enhancement is performed on the visible light data to obtain the multiple visible light data under different ambient light conditions at each of the positions.

4. The method for training a positioning model according to any one of claims 1 to 3, characterized in that: The taking each visible light data, the position information corresponding to each visible light data, and the ambient light identifier of the ambient light corresponding to each visible light data as training data samples includes: Separating the light data corresponding to the different visible light sources in the visible light data to obtain first RSS data samples corresponding to each visible light data, wherein the first RSS data samples include a plurality of first RSS data corresponding to different visible light sources; Each of the first RSS data samples, the position information corresponding to each of the first RSS data samples, and the ambient light identifier corresponding to each of the first RSS data samples are used as training data samples.

5. The method for training a model for positioning according to claim 4, characterized in that: The first neural network model includes a first neural network submodule and a second neural network submodule; The step of inputting each of the visible light data, the position information corresponding to each of the visible light data, and the ambient light identifier of the ambient light corresponding to each of the visible light data as training data samples into a first neural network model includes: inputting each of the first RSS data samples, the location information corresponding to each of the first RSS data samples, and the ambient light identifier corresponding to each of the first RSS data samples as training data samples into the first neural network submodule, wherein the first neural network submodule takes RSS data from different visible light sources under any ambient light at any position as input and takes weights corresponding to each of the RSS data as output; Weighting the RSS data in the first RSS data sample based on the weight corresponding to each RSS data in the first RSS data sample output by the first neural network submodule to obtain a second RSS data sample; Each of the second RSS data samples, the location information corresponding to each of the second RSS data samples, and the ambient light identifier corresponding to each of the second RSS data samples are input into the second neural network submodule.

6. The method for training a model for positioning according to claim 5, characterized in that: The first neural network submodule includes an attention mechanism.

7. The method for training a model for positioning according to claim 5, characterized in that: The first neural network submodule is a multi-layer perceptron with an attention mechanism.

8. The method for training a positioning model according to claim 5, wherein: The second neural network submodule includes a long short-term memory network layer and a fully connected layer.

9. The method for training a positioning model according to claim 1, wherein: Also includes: Based on the predicted positioning data output by the second neural network model and the actual position data corresponding to the predicted positioning data, the parameters of the second neural network model are optimized.

10. The method for training a positioning model according to claim 9, wherein: Optimizing the parameters of the second neural network model includes: Based on the predicted positioning data and the actual position data, determining an error between the predicted positioning data and the actual position data using a loss function; Optimize parameters of the second neural network model based on the error.

11. The method for training a positioning model according to claim 10, wherein: The loss function is expressed by the following formula: in, β represents the regularization adjustment coefficient of the regularization term; N represents the number of visible light sources; d j It represents the distance between the position corresponding to the visible light data calculated based on the Lambertian model and the jth visible light source; RSS j Represents the light data corresponding to the jth visible light source; ε characterizes the channel noise, and the channel noise is Gaussian noise.

12. The method for training a positioning model according to any one of claims 1 to 3, characterized in that: At least some of the multiple positions correspond to the same motion trajectory.

13. A positioning method, characterized in that: include: Obtaining the collected measured visible light data; The measured visible light data is input into the second neural network model trained by the training method for the positioning model described in any one of claims 1 to 12, and the predicted position corresponding to the measured visible light data is output.

14. An electronic device, characterized in that: include: one or more processors; One or more memories; the one or more memories store one or more programs, and when the one or more programs are executed by the one or more processors, the electronic device executes the training method for the positioning model described in any one of claims 1 to 12 or the positioning method described in claim 13.

15. A computer-readable storage medium, characterized in that The storage medium stores instructions, which, when executed on a computer, enable the computer to execute the training method for a positioning model according to any one of claims 1 to 12 or the positioning method according to claim 13.

16. A computer program product, characterized in that The computer program product comprises instructions, which, when executed, cause a computer to perform the method for training a model for positioning according to any one of claims 1 to 12 or the positioning method according to claim 13 .

Images