An adaptive object positioning method and apparatus

Abstract

The application relates to the technical field of wireless sensing, and discloses a self-adaptive object positioning method and device, which comprises the following steps: acquiring first signal strength information of communication between a measured object and at least one first communication device; inputting all the first signal strength information into a preset information processing model set to obtain first attribute information of the measured object, wherein the first attribute information comprises position information, the preset information processing model set satisfies a preset model training convergence condition, and comprises a preset signal strength filtering model, a preset channel calculation model and a preset regression model; the preset signal strength filtering model is used for performing a preset filtering processing operation on all the first signal strength information; the preset channel calculation model is used for simulating a channel environment of communication between the measured object and each first communication device; and the preset regression model is used for simulating an environment of a position where each first communication device is located. It can be seen that the application can improve the accuracy of object positioning.

Description

Technical Field

[0001] This invention relates to the field of wireless sensing technology, and in particular to an adaptive object localization method and apparatus. Background Technology

[0002] With the continuous development of technology, radio frequency identification (RFID) technology can detect the location of more and more objects without positioning devices, improving the flexibility of users' perception of everything.

[0003] Existing radio frequency identification (RFID) technologies mostly use the three-point localization method to locate the position of the object under test. That is, the position of the object under test is determined by the positions of three known points. However, for the three-point localization method, the solution space is often a region rather than an accurate position. In order to further determine the accurate position and thus the position of the object under test, the centroid of the region is usually taken as the position of the object under test. This implicitly assumes that the three known points in the three-point localization method contribute equally to the position of the object under test, ignoring the differences between different known points, thereby reducing the accuracy of the position determination of the object under test.

[0004] Therefore, it is particularly important to propose a technical solution to improve the accuracy of object positioning. Summary of the Invention

[0005] This invention provides an adaptive object positioning method and apparatus, which can improve the accuracy of object positioning.

[0006] To address the aforementioned technical problems, the first aspect of this invention discloses an adaptive object localization method, the method comprising:

[0007] Acquire first signal strength information of communication between the object under test and at least one first communication device;

[0008] All the first signal strength information is input into a preset information processing model set to obtain the first attribute information of the object under test. The first attribute information includes location information. The preset information processing model set satisfies the preset model training convergence condition. The preset information processing model set includes a preset signal strength filtering model, a preset channel calculation model, and a preset regression model. The preset signal strength filtering model is used to perform preset filtering processing operations on all the first signal strength information. The preset channel calculation model is used to simulate the channel environment for communication between the object under test and each of the first communication devices. The preset regression model is used to simulate the environment of the location of each of the first communication devices.

[0009] As an optional implementation, in the first aspect of the present invention, the step of inputting all the first signal strength information into a preset information processing model set to obtain the first attribute information of the object under test includes:

[0010] All the first signal strength information is input into the preset signal strength filtering model to obtain the target signal strength information;

[0011] The target signal strength information is input into the preset channel calculation model to obtain the target distance information between the object under test and each of the first communication devices;

[0012] The target signal strength information and the target distance information are input into the preset regression model to obtain the first attribute information of the object under test.

[0013] As an optional implementation, in the first aspect of the present invention, the preset signal strength filtering model is:

[0014]

[0015] Among them, RSSI out The term "maxpooling" is used to represent the target signal strength information, and it represents the maximum pooling operation. The convolution kernel used to represent the preset signal strength filtering model, b CNN RSSI is used to represent the bias of the convolution kernel. in This is used to represent a first matrix, the first matrix including a second matrix composed of the first signal strength information obtained from communication between the measured object and each of the first communication devices, and η is used to represent the total number of the first communication devices, and, The second matrix used to represent the i-th first communication device, and RSSI im The m-th first signal strength information used to represent the communication between the measured object and the i-th first communication device.

[0016] As an optional implementation, in the first aspect of the present invention, the preset channel calculation model is:

[0017] y m =w2*tanh(w1*RSSI) out +b1)+b2

[0018] Among them, y mThe target distance information is represented by tanh(), which represents a nonlinear activation function used to perform a nonlinear transformation operation. w1 represents the first weight parameter of the preset channel calculation model, w2 represents the second weight parameter of the preset channel calculation model, b1 represents the first bias of the preset channel calculation model, and b2 represents the second bias of the preset channel calculation model.

[0019] As an optional implementation, in the first aspect of the present invention, the preset regression model is:

[0020] y out =w out *x+b out

[0021] Among them, y out w is used to represent the first attribute information of the object being measured. out b is used to represent the output weight parameters of the preset regression model. out Used to represent the output bias of the preset regression model, x = [RSSI] out ,y m ,r1,...r η ], and r η =(rx η ,ry η ) is used to represent the first true coordinates of the ηth first communication device in the first preset coordinate system, rx η Used to represent the x-coordinate of the ηth first communication device in the first preset coordinate system, ry η This is used to represent the ordinate of the ηth first communication device in the first preset coordinate system.

[0022] As an optional implementation, in the first aspect of the present invention, the training steps of the preset information processing model set include:

[0023] For each target object in the target environment, a target identifier for the target object is determined based on the second attribute information of the target object; a second signal strength information for communication between the target object and each second communication device is determined based on the target identifier; and the second signal strength information, the second attribute information of the target object, and the third attribute information of each second communication device are determined as the first information set of the target object.

[0024] For each of the first information sets, according to the preset processing conditions, the preprocessing operation corresponding to the preset processing conditions is performed on the first information set to obtain the second information set of the first information set;

[0025] Based on the preset cross-validation conditions, all the second information sets are divided into training sample datasets and test sample datasets;

[0026] According to the preset training conditions, each training sample data in the training sample dataset is input into the original information processing model set to obtain the first predicted attribute information of each training sample data and the preparatory information processing model set corresponding to the original information processing model set.

[0027] Based on all the first predicted attribute information, it is determined whether the preparatory information processing model set meets the preset verification conditions. When it is determined that the preparatory information processing model set meets the preset verification conditions, each test sample data in the test sample dataset is input into the preparatory information processing model set according to the preset verification conditions to obtain the second predicted attribute information of each test sample data.

[0028] Based on all the second prediction attribute information, determine whether the set of preliminary information processing models meets the preset model training convergence condition. When it is determined that the set of preliminary information processing models meets the preset model training convergence condition, the set of preliminary information processing models is determined as the preset information processing model set.

[0029] As an optional implementation, in the first aspect of the present invention, the preset verification condition and the preset model training convergence condition include the convergence of a preset loss function, wherein the preset loss function is:

[0030]

[0031] Where y represents the second true coordinates of the target object in the second preset coordinate system, and d represents the true communication distance between the target object and each of the second communication devices. my Used to represent the predicted distance information between the target object and each of the second communication devices, |w out |、|w2|、 For regularization parameters;

[0032] When the preset loss function is used to verify whether the set of preliminary information processing models meets the preset verification conditions, y outy Used to represent the first predicted attribute information;

[0033] When the preset loss function is used to verify whether the pre-set information processing model set satisfies the preset model training convergence condition, y outy Used to represent the second predicted attribute information.

[0034] A second aspect of the present invention discloses an adaptive object positioning device, the device comprising:

[0035] The acquisition module is used to acquire first signal strength information of communication between the object under test and at least one first communication device;

[0036] The processing module is used to input all the first signal strength information into a preset information processing model set to obtain the first attribute information of the object under test. The first attribute information includes location information. The preset information processing model set satisfies the preset model training convergence condition. The preset information processing model set includes a preset signal strength filtering model, a preset channel calculation model, and a preset regression model. The preset signal strength filtering model is used to perform preset filtering processing operations on all the first signal strength information. The preset channel calculation model is used to simulate the channel environment for communication between the object under test and each of the first communication devices. The preset regression model is used to simulate the environment of the location of each of the first communication devices.

[0037] As an optional implementation, in a second aspect of the present invention, the specific method by which the processing module inputs all the first signal strength information into a preset information processing model set to obtain the first attribute information of the object under test includes:

[0038] All the first signal strength information is input into the preset signal strength filtering model to obtain the target signal strength information;

[0039] The target signal strength information is input into the preset channel calculation model to obtain the target distance information between the object under test and each of the first communication devices;

[0040] The target signal strength information and the target distance information are input into the preset regression model to obtain the first attribute information of the object under test.

[0041] As an optional implementation, in the second aspect of the present invention, the preset signal strength filtering model is:

[0042]

[0043] Among them, RSSI out The term "maxpooling" is used to represent the target signal strength information, and it represents the maximum pooling operation. The convolution kernel used to represent the preset signal strength filtering model, b CNN RSSI is used to represent the bias of the convolution kernel. inThis is used to represent a first matrix, the first matrix including a second matrix composed of the first signal strength information obtained from communication between the measured object and each of the first communication devices, and η is used to represent the total number of the first communication devices, and, The second matrix used to represent the i-th first communication device, and RSSI im The m-th first signal strength information used to represent the communication between the measured object and the i-th first communication device.

[0044] As an optional implementation, in the second aspect of the present invention, the preset channel calculation model is:

[0045] y m =w2*tanh(w1*RSSI) out +b1)+b2

[0046] Among them, y m The target distance information is represented by tanh(), which represents a nonlinear activation function used to perform a nonlinear transformation operation. w1 represents the first weight parameter of the preset channel calculation model, w2 represents the second weight parameter of the preset channel calculation model, b1 represents the first bias of the preset channel calculation model, and b2 represents the second bias of the preset channel calculation model.

[0047] As an optional implementation, in a second aspect of the present invention, the preset regression model is:

[0048] y out =w out *x+b out

[0049] Among them, y out w is used to represent the first attribute information of the object being measured. out b is used to represent the output weight parameters of the preset regression model. out Used to represent the output bias of the preset regression model, x = [RSSI] out ,y m ,r1,...r η ], and r η =(rx η ,ry η ) is used to represent the first true coordinates of the ηth first communication device in the first preset coordinate system, rx η Used to represent the x-coordinate of the ηth first communication device in the first preset coordinate system, ry ηThis is used to represent the ordinate of the ηth first communication device in the first preset coordinate system.

[0050] As an optional implementation, in a second aspect of the present invention, the training steps of the preset information processing model set include:

[0051] For each target object in the target environment, a target identifier for the target object is determined based on the second attribute information of the target object; a second signal strength information for communication between the target object and each second communication device is determined based on the target identifier; and the second signal strength information, the second attribute information of the target object, and the third attribute information of each second communication device are determined as the first information set of the target object.

[0052] For each of the first information sets, according to the preset processing conditions, the preprocessing operation corresponding to the preset processing conditions is performed on the first information set to obtain the second information set of the first information set;

[0053] Based on the preset cross-validation conditions, all the second information sets are divided into training sample datasets and test sample datasets;

[0054] According to the preset training conditions, each training sample data in the training sample dataset is input into the original information processing model set to obtain the first predicted attribute information of each training sample data and the preparatory information processing model set corresponding to the original information processing model set.

[0055] Based on all the first predicted attribute information, it is determined whether the preparatory information processing model set meets the preset verification conditions. When it is determined that the preparatory information processing model set meets the preset verification conditions, each test sample data in the test sample dataset is input into the preparatory information processing model set according to the preset verification conditions to obtain the second predicted attribute information of each test sample data.

[0056] Based on all the second prediction attribute information, determine whether the set of preliminary information processing models meets the preset model training convergence condition. When it is determined that the set of preliminary information processing models meets the preset model training convergence condition, the set of preliminary information processing models is determined as the preset information processing model set.

[0057] As an optional implementation, in the second aspect of the present invention, the preset verification condition and the preset model training convergence condition include the convergence of a preset loss function, wherein the preset loss function is:

[0058]

[0059] Where y represents the second true coordinates of the target object in the second preset coordinate system, and d represents the true communication distance between the target object and each of the second communication devices. my Used to represent the predicted distance information between the target object and each of the second communication devices, |w out |、|w2|、 For regularization parameters;

[0060] When the preset loss function is used to verify whether the set of preliminary information processing models meets the preset verification conditions, y outy Used to represent the first predicted attribute information;

[0061] When the preset loss function is used to verify whether the pre-set information processing model set satisfies the preset model training convergence condition, y outy Used to represent the second predicted attribute information.

[0062] A third aspect of the present invention discloses another adaptive object positioning device, the device comprising:

[0063] Memory containing executable program code;

[0064] A processor coupled to the memory;

[0065] The processor calls the executable program code stored in the memory to execute the adaptive object localization method disclosed in the first aspect of the present invention.

[0066] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute the adaptive object localization method disclosed in the first aspect of the present invention.

[0067] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0068] In this embodiment of the invention, first signal strength information of communication between the object under test and at least one first communication device is acquired; all first signal strength information is input into a preset information processing model set to obtain first attribute information of the object under test, including location information. The preset information processing model set satisfies preset model training convergence conditions. The preset information processing model set includes a preset signal strength filtering model, a preset channel calculation model, and a preset regression model. The preset signal strength filtering model is used to perform preset filtering processing operations on all first signal strength information. The preset channel calculation model is used to simulate the channel environment of communication between the object under test and each first communication device. The preset regression model is used to simulate the environment of the location of each first communication device. Therefore, implementing this invention can determine the location information of the object under test based on the acquired first signal strength information of communication between the object under test and at least one first communication device, and the preset signal strength filtering model, preset channel calculation model, and preset regression model in the preset information model set that satisfy the preset model training convergence conditions. This can fully combine the channel environment of communication between the object under test and each first communication device and the environment of the location of each first communication device, thereby improving the accuracy of object positioning. Attached Figure Description

[0069] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0070] Figure 1 This is a flowchart illustrating an adaptive object localization method disclosed in an embodiment of the present invention;

[0071] Figure 2 This is a schematic diagram of the structure of an adaptive object positioning device disclosed in an embodiment of the present invention;

[0072] Figure 3 This is a schematic diagram of the structure of another adaptive object positioning device disclosed in an embodiment of the present invention;

[0073] Figure 4 This is a schematic diagram of the structure of a preset information processing model set disclosed in an embodiment of the present invention;

[0074] Figure 5 This is a schematic diagram of a deployment scenario for a set of preset information processing models disclosed in an embodiment of the present invention. Detailed Implementation

[0075] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0077] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0078] This invention discloses an adaptive object localization method and apparatus. Based on the acquired signal strength information of the communication between the object under test and at least one first communication device, and a preset signal strength filtering model, a preset channel calculation model, and a preset regression model from a preset information model set that satisfy preset model training convergence conditions, the method determines the position information of the object under test. It effectively combines the channel environment of the communication between the object under test and each first communication device with the environment of the location of each first communication device, thereby improving the accuracy of object localization. Detailed descriptions follow.

[0079] Example 1

[0080] Please see Figure 1 , Figure 1 This is a flowchart illustrating an adaptive object localization method disclosed in an embodiment of the present invention. Figure 1The described adaptive object localization method can be applied to positioning devices, satellite devices, and intelligent devices associated with positioning devices and / or satellite devices. "Associated" can refer to electrical connection, wireless connection, indirect connection, or accessory. The aforementioned intelligent devices can include, but are not limited to, one or more of cloud devices, edge computing devices, relay devices, base station devices, smart home devices, urban management devices, and intelligent connected devices. This invention does not limit the scope of these devices. Figure 1 As shown, the adaptive object localization method may include the following operations:

[0081] 101. Obtain first signal strength information for communication between the object under test and at least one first communication device.

[0082] In this embodiment of the invention, optionally, the acquisition of the first signal strength information of the communication between the object under test and at least one first communication device may include the following operations:

[0083] For each first communication device, the identification information of the object under test is determined; based on the identification information, a first sensing signal is determined and sent to the location of the identification information, and a second sensing signal is received, the second sensing signal being a feedback signal of the first sensing signal; based on the first sensing signal and the second sensing signal, the first signal strength information of the communication between the object under test and the first communication device is determined.

[0084] In this optional embodiment, the aforementioned identification information may be at least one of the following: electronic tag, trademark of the object being tested, size, shape, material, weight, color, smoothness, light transmittance, and radioactivity.

[0085] In this optional embodiment, the location of the aforementioned identification information may or may not include the location of the object being measured relative to the first communication device.

[0086] As can be seen, implementing this optional embodiment can determine the first sensing signal based on the identification information of the object being measured, thereby sending the first sensing signal to the location of the identification information and receiving the second sensing signal. Then, based on the first sensing signal and the second sensing signal, the first signal strength information between the object being measured and the first communication device can be determined. This can improve the accuracy of determining the first signal strength information and further enhance the flexibility of determining the first signal strength information.

[0087] 102. Input all first signal strength information into the preset information processing model set to obtain the first attribute information of the object under test. The first attribute information includes location information. The preset information processing model set satisfies the preset model training convergence condition. The preset information processing model set includes a preset signal strength filtering model, a preset channel calculation model, and a preset regression model. The preset signal strength filtering model is used to perform preset filtering processing operations on all first signal strength information. The preset channel calculation model is used to simulate the channel environment of communication between the object under test and each first communication device. The preset regression model is used to simulate the environment of the location of each first communication device.

[0088] In this embodiment of the invention, the first attribute information may further include the identification information.

[0089] In this embodiment of the invention, optionally, the above-mentioned preset signal strength filtering model is as follows: Figure 4 As shown, it can include convolutional layers and max pooling layers. Figure 4 In this context, the first communication device is limited to base station equipment; however, it may also include, but is not limited to, mobile phones, PCs, game consoles, and in-vehicle intelligent connected devices. Figure 4 In the embodiments of the present invention, all first signal strength information corresponds to Figure 4 The "preprocessed location dataset" in this embodiment of the invention corresponds to the preset signal strength filtering model. Figure 4 The RSSI filtering network in the present invention corresponds to the preset channel calculation model in this embodiment. Figure 4 The channel transmission calculation model in the present invention corresponds to the preset regression model in this embodiment. Figure 4 The linear regression model in [the context].

[0090] As can be seen, implementing the embodiments of the present invention can determine the location information of the object under test based on the first signal strength information of the communication between the object under test and at least one first communication device, and the preset signal strength filtering model, preset channel calculation model and preset regression model in the preset information model set that meet the preset model training convergence conditions. It can fully combine the channel environment of the communication between the object under test and each first communication device and the environment of the location of each first communication device, thereby improving the accuracy of the object under test location.

[0091] In this embodiment of the invention, as an optional implementation, the above-mentioned inputting all the first signal strength information into a preset information processing model set to obtain the first attribute information of the object under test may include the following operations:

[0092] All the first signal strength information is input into the preset signal strength filtering model to obtain the target signal strength information.

[0093] The target signal strength information is input into the preset channel calculation model to obtain the target distance information between the object under test and each first communication device.

[0094] The target signal strength information and target distance information are input into a preset regression model to obtain the first attribute information of the object being measured.

[0095] As can be seen, implementing this optional embodiment discloses a model architecture of a preset information processing model set. By inputting all first signal strength information into a preset signal strength filtering model, target signal strength information can be obtained. Then, the target signal strength information is input into a preset channel calculation model to obtain target distance information between the measured object and each first communication device. Finally, the target signal strength information and target distance information are input into a preset regression model to obtain the first attribute information of the measured object. Through a specific model architecture, combining the different functions of the preset signal strength filtering model, the preset channel calculation model, and the preset regression model, the accuracy of object positioning can be further improved.

[0096] In this optional embodiment, as an optional implementation method, the above-mentioned preset signal strength filtering model is:

[0097]

[0098] Among them, RSSI out The term "maxpooling" is used to represent the target signal strength information, and it represents the maximum pooling operation. The convolution kernel used to represent the preset signal strength filtering model, b CNN RSSI is used to represent the bias of the convolution kernel. in Used to represent a first matrix, the first matrix including a second matrix composed of first signal strength information obtained from communication between the measured object and each first communication device, and η is used to represent the total number of the first communication devices, and, The second matrix is ​​used to represent the i-th first communication device, and RSSI im This is used to represent the m-th first signal strength information used in communication between the measured object and the i-th first communication device.

[0099] As can be seen, the implementation of this optional embodiment discloses a model architecture for a preset signal strength filtering model, which can improve the accuracy of determining the target signal strength information by specifically performing convolution and max pooling operations on the signal strength information. This is beneficial to improving the accuracy of determining the target distance information and the position information of the measured object.

[0100] In this optional embodiment, as another optional implementation, the aforementioned preset channel calculation model is:

[0101] y m =w2*tanh(w1*RSSI) out +b1)+b2

[0102] Among them, y m The function is used to represent target distance information, tanh() is used to represent a nonlinear activation function, the nonlinear activation function is used to perform nonlinear transformation operations, w1 is used to represent the first weight parameter of the preset channel calculation model, w2 is used to represent the second weight parameter of the preset channel calculation model, b1 is used to represent the first bias of the preset channel calculation model, and b2 is used to represent the second bias of the preset channel calculation model.

[0103] As can be seen, the implementation of this optional embodiment discloses a model architecture of a preset channel calculation model, which can simulate the channel environment of communication between the object under test and each first communication device through the trained preset channel calculation model, thereby comprehensively considering the impact of channel interference on the positioning of the object under test, improving the accuracy of target distance information determination, and further facilitating the combination of target signal strength information and preset regression model to improve the accuracy of the position information determination of the object under test.

[0104] In this optional embodiment, as yet another optional implementation, the aforementioned preset regression model is:

[0105] y out =w out *x+b out

[0106] Among them, y out w is used to represent the first attribute information of the object being measured. out b is used to represent the output weight parameters of the preset regression model. out Used to represent the output bias of the preset regression model, x = [RSSI] out ,y m ,r1,...r η ], and r η =(rx η ,ry η ) is used to represent the first true coordinates of the ηth first communication device in the first preset coordinate system, rx η Used to represent the x-coordinate of the ηth first communication device in the first preset coordinate system, ry η Used to represent the ordinate of the ηth first communication device in the first preset coordinate system.

[0107] As can be seen, the implementation of this optional embodiment discloses a model architecture of a preset regression model, which can simulate the environment of the location of each first communication device through the trained preset regression model, thereby adaptively determining the contribution weight of different first communication devices to the measured object, and improving the accuracy of determining the location information of the measured object by combining the target signal strength information and target distance information mentioned above.

[0108] In an optional embodiment, the training steps of the aforementioned preset information processing model set include:

[0109] For each target object in the target environment, the target identifier of the target object is determined based on the second attribute information of the target object; the second signal strength information of the communication between the target object and each second communication device is determined based on the target identifier; the second signal strength information, the second attribute information of the target object and the third attribute information of each second communication device are determined as the first information set of the target object.

[0110] For each first information set, according to preset processing conditions, perform the preprocessing operation corresponding to the preset processing conditions on the first information set to obtain the second information set of the first information set.

[0111] Based on the preset cross-validation conditions, all second information sets are divided into training sample datasets and test sample datasets.

[0112] Based on the preset training conditions, each training sample data in the training sample dataset is input into the original information processing model set to obtain the first predicted attribute information of each training sample data and the preparatory information processing model set corresponding to the original information processing model set.

[0113] Based on all the first predicted attribute information, determine whether the preparatory information processing model set meets the preset verification conditions. When it is determined that the preparatory information processing model set meets the preset verification conditions, then according to the preset verification conditions, input each test sample data in the test sample dataset into the preparatory information processing model set to obtain the second predicted attribute information of each test sample data.

[0114] Based on all the second prediction attribute information, determine whether the set of preliminary information processing models meets the preset model training convergence condition. If it is determined that the set of preliminary information processing models meets the preset model training convergence condition, then the set of preliminary information processing models is determined as the preset information processing model set.

[0115] In this optional embodiment, the first information set includes, but is not limited to, the location information of the second communication device, the location information of the target object, the location information of the target object relative to the second communication device, the target identifier, and the second signal strength information. Specifically, it can be obtained by performing a preset calculation process on the second signal strength information, the second attribute information of the target object, and the third attribute information of each second communication device.

[0116] Optionally, the above preprocessing operations may include, but are not limited to, at least one of the following: data normalization, deletion of missing values, etc.

[0117] Further optionally, the aforementioned preset cross-validation conditions may include, but are not limited to, at least one of 5-fold cross-validation and 10-fold cross-validation. Specifically, the number of cross-validations can be determined according to the actual application scenario, and this embodiment of the invention does not impose specific limitations on this.

[0118] As can be seen, implementing this optional embodiment discloses a training step for a preset information processing model set, which can improve the accuracy of the preset signal strength filtering model, preset channel calculation model and preset regression model in the preset information processing model set in achieving their respective functions, thereby improving the accuracy of the object being measured.

[0119] In another optional embodiment, the aforementioned preset verification conditions and preset model training convergence conditions include preset loss function convergence, wherein the preset loss function is:

[0120]

[0121] Where y represents the second true coordinates of the target object in the second preset coordinate system, and d represents the true communication distance between the target object and each second communication device. my Used to represent the predicted distance information between the target object and each second communication device, |w out |、|w2|、 This is the regularization parameter.

[0122] When the preset loss function is used to verify whether the set of preliminary information processing models meets the preset verification conditions, y outy Used to represent the first predicted attribute information.

[0123] When the preset loss function is used to verify whether the set of preliminary information processing models meets the preset model training convergence condition, y outy Used to represent the second predicted attribute information.

[0124] As can be seen, implementing this optional embodiment can determine the preset loss function through the preset information processing model set, thereby verifying whether the preset information processing model set meets the preset verification conditions and whether the preset information processing model set meets the preset model training convergence conditions, thereby further improving the accuracy of the preset signal strength filtering model, preset channel calculation model and preset regression model in the preset information processing model set to achieve the corresponding functions and improving the accuracy of the object positioning.

[0125] In yet another alternative embodiment, such as Figure 5 As shown, the preset information processing model set in the embodiments of the present invention corresponds to Figure 5 The server's positioning model, i.e., the aforementioned preset information processing model set, can be deployed on the server. Optionally, the server is wired or wirelessly connected to the first communication device, thereby achieving wireless channel communication through the aforementioned identification information. The aforementioned identification information corresponds to... Figure 5 RFID tags in the system.

[0126] Example 2

[0127] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an adaptive object positioning device disclosed in an embodiment of the present invention. Figure 2 The described adaptive object positioning device can be applied to positioning equipment, satellite equipment, and intelligent devices associated with positioning equipment and / or satellite equipment. "Associated" can refer to electrical connection, wireless connection, indirect connection, or accessory. The aforementioned intelligent devices may include, but are not limited to, one or more of cloud devices, edge computing devices, relay devices, base station devices, smart home devices, urban management devices, and intelligent connected devices. This invention does not limit the scope of these devices. Figure 2 As shown, the adaptive object positioning device may include:

[0128] The acquisition module 201 is used to acquire first signal strength information of communication between the object under test and at least one first communication device.

[0129] The processing module 202 is used to input all the first signal strength information into a preset information processing model set to obtain the first attribute information of the object under test. The first attribute information includes position information. The preset information processing model set satisfies the preset model training convergence condition. The preset information processing model set includes a preset signal strength filtering model, a preset channel calculation model, and a preset regression model. The preset signal strength filtering model is used to perform preset filtering processing operations on all the first signal strength information. The preset channel calculation model is used to simulate the channel environment of communication between the object under test and each first communication device. The preset regression model is used to simulate the environment of the location of each first communication device.

[0130] As can be seen, implementing the embodiments of the present invention can determine the location information of the object under test based on the first signal strength information of the communication between the object under test and at least one first communication device, and the preset signal strength filtering model, preset channel calculation model and preset regression model in the preset information model set that meet the preset model training convergence conditions. It can fully combine the channel environment of the communication between the object under test and each first communication device and the environment of the location of each first communication device, thereby improving the accuracy of the object under test location.

[0131] In this embodiment of the invention, as an optional implementation, the specific method by which the processing module 202 inputs all the first signal strength information into a preset information processing model set to obtain the first attribute information of the object under test includes:

[0132] All the first signal strength information is input into the preset signal strength filtering model to obtain the target signal strength information.

[0133] The target signal strength information is input into the preset channel calculation model to obtain the target distance information between the object under test and each first communication device.

[0134] The target signal strength information and target distance information are input into a preset regression model to obtain the first attribute information of the object being measured.

[0135] As can be seen, implementing this optional embodiment discloses a model architecture of a preset information processing model set. By inputting all first signal strength information into a preset signal strength filtering model, target signal strength information can be obtained. Then, the target signal strength information is input into a preset channel calculation model to obtain target distance information between the measured object and each first communication device. Finally, the target signal strength information and target distance information are input into a preset regression model to obtain the first attribute information of the measured object. Through a specific model architecture, combining the different functions of the preset signal strength filtering model, the preset channel calculation model, and the preset regression model, the accuracy of object positioning can be further improved.

[0136] In this optional embodiment, as an optional implementation method, the above-mentioned preset signal strength filtering model is:

[0137]

[0138] Among them, RSSI out The term "maxpooling" is used to represent the target signal strength information, and it represents the maximum pooling operation. The convolution kernel used to represent the preset signal strength filtering model, b CNN RSSI is used to represent the bias of the convolution kernel. in Used to represent a first matrix, the first matrix including a second matrix composed of first signal strength information obtained from communication between the measured object and each first communication device, and η is used to represent the total number of the first communication devices, and, The second matrix is ​​used to represent the i-th first communication device, and RSSI im This is used to represent the m-th first signal strength information used in communication between the measured object and the i-th first communication device.

[0139] As can be seen, the implementation of this optional embodiment discloses a model architecture for a preset signal strength filtering model, which can improve the accuracy of determining the target signal strength information by specifically performing convolution and max pooling operations on the signal strength information. This is beneficial to improving the accuracy of determining the target distance information and the position information of the measured object.

[0140] In this optional embodiment, as another optional implementation, the aforementioned preset channel calculation model is:

[0141] y m =w2*tanh(w1*RSSI) out +b1)+b2

[0142] Among them, y m The function is used to represent target distance information, tanh() is used to represent a nonlinear activation function, the nonlinear activation function is used to perform nonlinear transformation operations, w1 is used to represent the first weight parameter of the preset channel calculation model, w2 is used to represent the second weight parameter of the preset channel calculation model, b1 is used to represent the first bias of the preset channel calculation model, and b2 is used to represent the second bias of the preset channel calculation model.

[0143] As can be seen, the implementation of this optional embodiment discloses a model architecture of a preset channel calculation model, which can simulate the channel environment of communication between the object under test and each first communication device through the trained preset channel calculation model, thereby comprehensively considering the impact of channel interference on the positioning of the object under test, improving the accuracy of target distance information determination, and further facilitating the combination of target signal strength information and preset regression model to improve the accuracy of the position information determination of the object under test.

[0144] In this optional embodiment, as yet another optional implementation, the aforementioned preset regression model is:

[0145] y out =w out *x+b out

[0146] Among them, y out w is used to represent the first attribute information of the object being measured. out b is used to represent the output weight parameters of the preset regression model. out Used to represent the output bias of the preset regression model, x = [RSSI] out ,y m ,r1,...r η ], and r η =(rx η ,ry η ) is used to represent the first true coordinates of the ηth first communication device in the first preset coordinate system, rx η Used to represent the x-coordinate of the ηth first communication device in the first preset coordinate system, ry η Used to represent the ordinate of the ηth first communication device in the first preset coordinate system.

[0147] As can be seen, the implementation of this optional embodiment discloses a model architecture of a preset regression model, which can simulate the environment of the location of each first communication device through the trained preset regression model, thereby adaptively determining the contribution weight of different first communication devices to the measured object, and improving the accuracy of determining the location information of the measured object by combining the target signal strength information and target distance information mentioned above.

[0148] In an optional embodiment, the training steps of the aforementioned preset information processing model set include:

[0149] For each target object in the target environment, the target identifier of the target object is determined based on the second attribute information of the target object; the second signal strength information of the communication between the target object and each second communication device is determined based on the target identifier; the second signal strength information, the second attribute information of the target object and the third attribute information of each second communication device are determined as the first information set of the target object.

[0150] For each first information set, according to preset processing conditions, perform the preprocessing operation corresponding to the preset processing conditions on the first information set to obtain the second information set of the first information set.

[0151] Based on the preset cross-validation conditions, all second information sets are divided into training sample datasets and test sample datasets.

[0152] Based on the preset training conditions, each training sample data in the training sample dataset is input into the original information processing model set to obtain the first predicted attribute information of each training sample data and the preparatory information processing model set corresponding to the original information processing model set.

[0153] Based on all the first predicted attribute information, determine whether the preparatory information processing model set meets the preset verification conditions. When it is determined that the preparatory information processing model set meets the preset verification conditions, then according to the preset verification conditions, input each test sample data in the test sample dataset into the preparatory information processing model set to obtain the second predicted attribute information of each test sample data.

[0154] Based on all the second prediction attribute information, determine whether the set of preliminary information processing models meets the preset model training convergence condition. If it is determined that the set of preliminary information processing models meets the preset model training convergence condition, then the set of preliminary information processing models is determined as the preset information processing model set.

[0155] As can be seen, implementing this optional embodiment discloses a training step for a preset information processing model set, which can improve the accuracy of the preset signal strength filtering model, preset channel calculation model and preset regression model in the preset information processing model set in achieving their respective functions, thereby improving the accuracy of the object being measured.

[0156] In another optional embodiment, the aforementioned preset verification conditions and preset model training convergence conditions include preset loss function convergence, wherein the preset loss function is:

[0157]

[0158] Where y represents the second true coordinates of the target object in the second preset coordinate system, and d represents the true communication distance between the target object and each second communication device. my Used to represent the predicted distance information between the target object and each second communication device, |w out |、|w2|、 This is the regularization parameter.

[0159] When the preset loss function is used to verify whether the set of preliminary information processing models meets the preset verification conditions, y outy Used to represent the first predicted attribute information.

[0160] When the preset loss function is used to verify whether the set of preliminary information processing models meets the preset model training convergence condition, y outy Used to represent the second predicted attribute information.

[0161] As can be seen, implementing this optional embodiment can determine the preset loss function through the preset information processing model set, thereby verifying whether the preset information processing model set meets the preset verification conditions and whether the preset information processing model set meets the preset model training convergence conditions, thereby further improving the accuracy of the preset signal strength filtering model, preset channel calculation model and preset regression model in the preset information processing model set to achieve the corresponding functions and improving the accuracy of the object positioning.

[0162] Example 3

[0163] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of another adaptive object positioning device disclosed in an embodiment of the present invention. For example... Figure 3 As shown, the adaptive object positioning device may include:

[0164] Memory 301 that stores executable program code.

[0165] Processor 302 coupled to memory 301.

[0166] The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the adaptive object localization method described in Embodiment 1 of the present invention.

[0167] Example 4

[0168] This invention discloses a computer storage medium storing computer instructions, which, when invoked, execute the steps in the adaptive object localization method described in Embodiment 1 of this invention.

[0169] Example 5

[0170] This invention discloses a computer program product, which includes a non-transitory computer storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the adaptive object localization method described in Embodiment 1.

[0171] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0172] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0173] Finally, it should be noted that the adaptive object positioning method and apparatus disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of adaptive object positioning, characterized by, The method comprises: obtaining first signal strength information of communication between a measured object and at least one first communication device; inputting all the first signal strength information into a preset information processing model set to obtain first attribute information of the measured object, the first attribute information comprising position information, the preset information processing model set satisfying a preset model training convergence condition, the preset information processing model set comprising a preset signal strength filtering model, a preset channel calculation model and a preset regression model, the preset signal strength filtering model being used for performing a preset filtering processing operation on all the first signal strength information, the preset channel calculation model being used for simulating a channel environment of communication between the measured object and each first communication device, and the preset regression model being used for simulating an environment of a position where each first communication device is located; and the inputting all the first signal strength information into the preset information processing model set to obtain the first attribute information of the measured object comprises: inputting all the first signal strength information into the preset signal strength filtering model to obtain target signal strength information; inputting the target signal strength information into the preset channel calculation model to obtain target distance information of the measured object and each first communication device; and inputting the target signal strength information and the target distance information into the preset regression model to obtain the first attribute information of the measured object.

2. The adaptive object positioning method of claim 1, wherein, The preset signal strength filtering model is: wherein, for representing the target signal strength information, for representing a max-pooling operation, for representing a convolution kernel of the preset signal strength filtering model, for representing a bias of the convolution kernel, for representing a first matrix, the first matrix comprising a second matrix composed of the first signal strength information obtained by communication between the measured object and each of the first communication devices, and , for representing a total number of the first communication devices, and, for representing the second matrix of the first communication device, , , for representing the first signal strength information obtained by communication between the measured object and the first communication device, , .

3. The adaptive object positioning method of claim 2, wherein, The preset channel calculation model is: wherein, for representing the target distance information, for representing a nonlinear activation function, the nonlinear activation function being used to perform a nonlinear transformation operation, for representing a first weight parameter of the preset channel calculation model, for representing a second weight parameter of the preset channel calculation model, for representing a first bias of the preset channel calculation model, for representing a second bias of the preset channel calculation model.

4. The adaptive object positioning method of claim 3, wherein, The preset regression model is: in, The first attribute information used to represent the object being measured. Used to represent the output weight parameters of the preset regression model Used to represent the output bias of the preset regression model. ,as well as, Used to represent the first preset coordinate system The first real coordinates of the first communication device. Used to represent the first preset coordinate system. The x-coordinate of the first communication device Used to represent the first preset coordinate system. The vertical coordinate of the first communication device.

5. The adaptive object positioning method according to any one of claims 1-4, characterized in that, The training step of the preset information processing model set comprises: for each target object in a target environment, determining a target identifier of the target object according to second attribute information of the target object, determining second signal strength information of communication between the target object and each second communication device according to the target identifier, and determining the second signal strength information, the second attribute information of the target object and third attribute information of each second communication device as a first information set of the target object; for each first information set, performing a pre-processing operation corresponding to a preset processing condition on the first information set according to the preset processing condition to obtain a second information set of the first information set; dividing all the second information sets into a training sample data set and a test sample data set according to a preset cross-validation condition; inputting each training sample data in the training sample data set into an original information processing model set according to a preset training condition to obtain first predicted attribute information of each training sample data and a preliminary information processing model set corresponding to the original information processing model set; and training the preliminary information processing model set according to a preset training condition to obtain the preset information processing model set. According to all the first predicted attribute information, it is judged whether the preliminary information processing model set meets a preset verification condition. When it is judged that the preliminary information processing model set meets the preset verification condition, each test sample data in the test sample data set is input into the preliminary information processing model set according to the preset verification condition, and second predicted attribute information of each test sample data is obtained. According to all the second predicted attribute information, it is judged whether the preliminary information processing model set meets the preset model training convergence condition. When it is judged that the preliminary information processing model set meets the preset model training convergence condition, the preliminary information processing model set is determined as the preset information processing model set.

6. The adaptive object positioning method of claim 5, wherein, The preset verification condition and the preset model training convergence condition include preset loss function convergence, and the preset loss function is: wherein, a second real coordinate of the target object in a second preset coordinate system, a real distance between the target object and each of the second communication devices, predicted distance information of the target object and each of the second communication devices, is a regularization parameter; When the preset loss function is used to verify whether the preliminary information processing model set meets a preset verification condition, for representing the first predicted attribute information; When the preset loss function is used to verify whether the preliminary information processing model set meets a preset model training convergence condition, for representing the second predicted attribute information.

7. An adaptive object positioning device, characterized by The device is used to perform the adaptive object positioning method in any one of claims 1-6, and the device includes: The acquisition module is configured to acquire first signal strength information of communication between a measured object and at least one first communication device. The processing module is configured to input all the first signal strength information into a preset information processing model set to obtain first attribute information of the measured object, the first attribute information including position information, the preset information processing model set meeting a preset model training convergence condition, the preset information processing model set including a preset signal strength filtering model, a preset channel calculation model, and a preset regression model, the preset signal strength filtering model being used to perform a preset filtering processing operation on all the first signal strength information, the preset channel calculation model being used to simulate a channel environment of communication between the measured object and each first communication device, and the preset regression model being used to simulate an environment of a position where each first communication device is located.

8. An adaptive object positioning device, characterized by The device includes: a memory storing executable program codes; a processor coupled with the memory; The processor invokes the executable program codes stored in the memory to perform the adaptive object positioning method in any one of claims 1-6.

9. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, and the computer instructions are used to perform the adaptive object positioning method in any one of claims 1-6 when invoked.

Images