Positioning method and device, vehicle and storage medium

Abstract

The invention provides a positioning method and device, a vehicle and a readable storage medium. The method comprises the following steps: acquiring a first data set, wherein the first data set is online data information measured in a first use scene; the first data set is input into a first neural network model to obtain a second data set, and the first neural network model is used for performing noise reduction processing on the online data information; and inputting the second data set into a second neural network model to obtain first prediction information of the positioning result, the second neural network model being an offline training model. According to the embodiment of the invention, on one hand, noise reduction processing is performed on the positioning parameter information based on the neural network model, so that the positioning accuracy can be improved; and on the other hand, the neural network model can continuously and automatically learn and adjust the weight through training to adapt to different use scenes and conditions, so that the noise reduction effect is improved, and the positioning accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a positioning method, device, vehicle and readable storage medium. Background Art

[0002] It has become a trend to use contactless devices (such as digital keys) to control vehicles without feeling. The device is connected to the vehicle via Bluetooth or other wireless methods. After the vehicle completes the necessary safety certification, the vehicle will track the position changes of the device in real time and trigger the vehicle control action when the relative position of the device and the vehicle meets the established rules.

[0003] A necessary condition for the sensorless control function is wireless positioning between the vehicle and the device, and the accuracy of positioning will determine the effect of sensorless control. The basic principle of wireless positioning is to obtain relevant positioning parameter information through measurement and estimate the location of the target source using the corresponding algorithm. However, in real-world usage scenarios, there are many factors that affect positioning accuracy and they are prone to change, making it impossible to guarantee the accuracy of wireless positioning. Summary of the invention

[0004] The embodiments of the present application provide a positioning method, device, vehicle, and readable storage medium, which are introduced below from the following aspects.

[0005] In a first aspect, a positioning method is provided, the method comprising: acquiring a first data set, the first data set being online data information measured in a first usage scenario; inputting the first data set into a first neural network model to obtain a second data set, the first neural network model being used to perform noise reduction processing on the online data information; inputting the second data set into a second neural network model to obtain first prediction information of a positioning result, the second neural network model being an offline training model.

[0006] In some implementations, the first neural network model is trained online based on the first data set.

[0007] In some implementations, the second neural network model is trained offline based on a third data set, where the third data set is offline data information measured in multiple usage scenarios, and / or the second neural network model is trained online based on the second data set.

[0008] In some implementations, inputting a first data set into a first neural network model to obtain a second data set includes: inputting the first data set into the first neural network model to obtain a fourth data set; and filtering the first data values ​​in the fourth data set to obtain the second data set.

[0009] In some implementations, the first data value is received signal strength indication RSSI data and / or distance information.

[0010] In some implementations, the method further includes: filtering the first prediction information based on a filtering model to obtain second prediction information.

[0011] In some implementations, the first data set includes RSSI data and coordinate information, and / or the first data set includes distance information and coordinate information.

[0012] In a second aspect, a positioning device is provided, which includes: an acquisition unit, used to acquire a first data set, the first data set being online data information measured in a first usage scenario; a first processing unit, used to input the first data set into a first neural network model to obtain a second data set, the first neural network model being used to perform noise reduction processing on the online data information; a second processing unit, used to input the second data set into a second neural network model to obtain first prediction information of a positioning result, the second neural network model being an offline training model.

[0013] In a third aspect, a vehicle is provided, comprising: a memory and a processor, wherein the memory is used to store program codes, and the processor is used to call the program codes in the memory so that the vehicle executes the method in the first aspect.

[0014] According to a fourth aspect, a computer program product is provided, the computer program product comprising: a computer program code, when the computer program code is run on a computer, the computer executes the method in the first aspect.

[0015] In a fifth aspect, a readable storage medium is provided, which stores a program code. When the computer program code runs on a computer, the computer executes the method in the first aspect.

[0016] In the related art, the positioning parameter information of the device obtained in the use environment is directly input into the positioning model for positioning. Due to the influence of various factors in the environment, these positioning parameter information will be mixed with noise, which will then be introduced into the positioning model, resulting in low positioning accuracy.

[0017] In the embodiment of the present application, on the one hand, after obtaining the positioning parameter information, the positioning parameter information is first subjected to noise reduction processing based on the neural network model, and then positioning is performed based on the positioning parameter information after the noise reduction processing, thereby helping to improve the accuracy of positioning. On the other hand, the neural network model can continuously learn and adjust the weights automatically through training to adapt to different usage scenarios and conditions, thereby improving the effect of noise reduction processing, thereby helping to improve the accuracy of positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flowchart of a positioning method provided in an embodiment of the present application.

[0019] Figure 2 A flowchart of a positioning method provided in another embodiment of the present application.

[0020] Figure 3 A flowchart of an implementation method of positioning calibration provided in an embodiment of the present application.

[0021] Figure 4 A schematic diagram of the structure of a positioning device provided in an embodiment of the present application.

[0022] Figure 5 A schematic diagram of the structure of a vehicle provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments.

[0024] It has become a trend to use contactless devices (such as digital keys) to control vehicles without feeling. The device connects to the vehicle via Bluetooth or other wireless methods. After the vehicle completes the necessary safety authentication, the vehicle will track the location changes of the device in real time and trigger vehicle control actions when the relative positions of the device and the vehicle meet the established rules. Vehicle control actions include, for example, the vehicle flashing its headlights to welcome the device when it approaches the vehicle, unlocking the doors when the device approaches the vehicle further, or locking the doors when the device moves away from the vehicle. These vehicle control actions do not require the user to operate the device or the vehicle, so they are called contactless control.

[0025] A necessary condition for the sensorless control function is wireless positioning between the vehicle and the device, and the accuracy of positioning will determine the effect of the sensorless control. The basic principle of wireless positioning is to obtain relevant positioning parameter information through measurement and use the corresponding algorithm to estimate the location of the target source. First, take Bluetooth positioning as an example to introduce the positioning method of related technologies to help understand.

[0026] Phase 1: Data Collection

[0027] Preparation: (1) Select a suitable Bluetooth module: Select a Bluetooth module with sufficient accuracy and stability; (2) Determine reference points: Select multiple reference points in the target area. These points should be easily identifiable and evenly distributed; (3) Install the Bluetooth module: Install the Bluetooth module (anchor point) at the reference point and ensure that it can work properly.

[0028] Data collection: (1) Move: The vehicle and each reference point move relative to each other; (2) Record RSSI values: At each reference point, use a handheld device or a receiver mounted on the vehicle to measure the received signal strength indication (RSSI) of the Bluetooth signal emitted by each anchor point; (3) Repeat measurement: In order to obtain more reliable data, each reference point should be measured multiple times and the RSSI value recorded each time.

[0029] Phase 2: Data processing

[0030] Build fingerprint map: Using the collected data, create a fingerprint map for each reference point, which is a vector containing the average RSSI values ​​of all anchor points.

[0031] Storing fingerprint images: storing the fingerprint images of each reference point in the database to form a fingerprint database.

[0032] The third stage: model training stage

[0033] Feature engineering: extracting useful features from fingerprint images.

[0034] Model selection: Select an appropriate machine learning model (such as K-nearest neighbor algorithm, support vector machine, decision tree, etc.) based on the characteristics of the problem.

[0035] Training model: Use the data in the fingerprint database to train the selected model so that the model can predict the closest true location based on the RSSI value.

[0036] Phase 4: Positioning

[0037] Real-time measurement; (1) Real-time RSSI measurement: When the vehicle is at an unknown position, measure the RSSI values ​​of all current anchor points; (2) Query the fingerprint database: compare these RSSI values ​​with the fingerprint images in the fingerprint database.

[0038] Position estimation: (1) Similarity matching: find the fingerprint closest to the current RSSI value, that is, the closest reference point; (2) Position estimation: estimate the vehicle's position based on the position of the closest reference point.

[0039] Multi-lateration positioning: If higher accuracy is required, a multi-lateration positioning algorithm, such as trilateration or triangulation, can be used to calculate a more precise location using the locations and RSSI values ​​of multiple reference points.

[0040] However, in real-world scenarios, there are many factors that affect positioning accuracy, and they are prone to change, making it impossible to guarantee the accuracy of wireless positioning. The following introduces various factors that affect positioning accuracy.

[0041] (1) Multipath effect: After electromagnetic waves propagate through different paths, the various component fields arrive at the receiving end at different times, and their respective phases are superimposed on each other, causing interference and signal distortion.

[0042] (2) Non-line-of-sight transmission: If blocked by obstacles, the signal strength will be greatly attenuated, and the signal transmission direction and rate may also change;

[0043] (3) Co-channel interference: If multiple devices operate in the same frequency band and their signals collide, it may cause signal distortion;

[0044] (4) Posture: Positioning terminals are generally handheld devices. The position, posture, and angle of the human body will have different degrees of influence on the signal;

[0045] (5) System level: RF performance, hardware measurement performance, deployment location of anchor points / base stations / beacons, the geometry of the anchor points / base stations / beacons, and the number of anchor points / base stations / beacons will all affect the received signal strength.

[0046] Due to the influence of the above factors, the complexity of the positioning algorithm is extremely high. In the related art, the positioning parameter information of the device obtained in the use environment is directly input into the positioning model for positioning. Due to the influence of various factors in the environment, these positioning parameter information will be mixed with noise, which will lead to the introduction of noise into the positioning model, resulting in low positioning accuracy.

[0047] In order to solve the above problems, in the embodiment of the present application, on the one hand, after obtaining the positioning parameter information, the positioning parameter information is first subjected to noise reduction processing based on the neural network model, and then the positioning parameter information after noise reduction processing is input into the positioning model for positioning, thereby helping to improve the accuracy of positioning. On the other hand, the neural network model can continuously learn and adjust the weights automatically through training to adapt to different usage scenarios and conditions, thereby improving the effect of noise reduction processing, thereby helping to improve the accuracy of positioning.

[0048] Combined with the following Figure 1 , the positioning method provided in the embodiment of the present application is introduced in detail. It should be understood that the positioning method in the embodiment of the present application can be applied to positioning calibration, and can also be applied to positioning in actual use of the vehicle.

[0049] like Figure 1 As shown, in step S110, a first data set is acquired.

[0050] The first data set may be data information measured in the first usage scenario. The data information in the first data set is related to positioning, or in other words, the data information in the first data set is positioning parameter information used for positioning. The positioning parameter information here may be fingerprint data related to the location.

[0051] The above data information may also be referred to as online data information. It should be understood that the online data information here may refer to data information collected in real time in the first usage scenario.

[0052] The first data set may include multiple data information. For example, the first data set may include RSSI data and coordinate information. For another example, the first data set may include distance information and coordinate information. For another example, the first data set may also include RSSI data and coordinate information, as well as distance information and coordinate information. There is a one-to-one correspondence between the above RSSI data and coordinate information (or distance information and coordinate information), that is, different RSSI data values ​​correspond to different coordinate information.

[0053] The above coordinate information may be the coordinate information of the device (such as a digital key) relative to the vehicle, or the above coordinate information may be the coordinate information of the vehicle relative to the device.

[0054] The use scenario mentioned in this article can be understood as the use environment of the vehicle or the surrounding environment of the vehicle. In some implementations, the use scenario can also be understood as the surrounding environment of the vehicle when it is parked. The use scenario can be the surrounding environment of the vehicle in the experimental site or the surrounding environment of the vehicle in actual use. For example, the first use scenario can be the surrounding environment of the vehicle in the garage site. The use scenario may include one or more factors mentioned above that affect the positioning accuracy (such as multipath effect).

[0055] In step S120, the first data set is input into the first neural network model to obtain a second data set.

[0056] The first neural network model is used to perform noise reduction processing on the data information in the first data set. The embodiment of the present application does not specifically limit the type of the first neural network model. For example, the first neural network model can be one of the following neural networks: a recurrent neural network, a long short-term memory network, a convolutional neural network, an attention mechanism neural network, and a generative adversarial network.

[0057] The second data set mentioned above is a data set obtained by performing noise reduction processing on the data information in the first data set. Noise reduction here may refer to removing some data values ​​that do not meet the requirements, such as data values ​​with large deviations.

[0058] As an example, taking Bluetooth positioning as an example, multiple sets of RSSI data are collected at multiple known locations (reference points). If there is a certain RSSI data exceeding a certain threshold among the multiple sets of RSSI data, the RSSI data can be considered abnormal and the value is removed from the data set.

[0059] In the embodiment of the present application, the data information in the first data set is subjected to noise reduction processing based on the first neural network model. The training process of the first neural network model is described in detail with examples below.

[0060] First, a first neural network model is established according to a certain neural network algorithm (such as a convolutional neural network), which includes selecting an appropriate loss function to measure the difference between the model prediction value and the true value. The loss function here can be, for example, a function related to RSSI data. Then, the first data set is divided into a training set, a validation set, and a test set, and the first neural network model is trained based on the training set to optimize the minimum loss function. Next, the performance of the first neural network model is evaluated based on the validation set to check whether the first neural network model is overfitting or underfitting, and the parameter weights in the first neural network model are adjusted according to the evaluation results. Finally, the capability of the first neural network model is evaluated based on an independent test set.

[0061] In order to further remove noise in the data information, in some implementations, after the first neural network model performs noise reduction processing on the first data set, the first data value in the online data information may be filtered to obtain the second data set. The first data value here may be RSSI data and / or distance information.

[0062] In some implementations, the first data value may be subjected to Gaussian filtering. Gaussian filtering can remove the influence of random noise. For example, RSSI data or distance information is usually affected by environmental factors, such as multipath effects, obstacle obstruction, etc., resulting in unstable signal strength. The basic idea of ​​Gaussian filtering is to assume that the noise obeys a normal distribution (Gaussian distribution), and by calculating the probability density of each data point, select those values ​​with higher probabilities as valid data. Gaussian filtering can further remove abnormal data information (i.e., noise), retain data information with a high probability of being correct, so as to improve the effect of removing noise, and help improve the accuracy of positioning.

[0063] In step S130, the second data set is input into the second neural network model to obtain first prediction information of the positioning result.

[0064] The second neural network model can use the data information in the second data set as input, and then predict the positioning result. The positioning result here can be the location information of the positioning, for example, the first prediction information can include coordinate information and / or positioning area information.

[0065] The second neural network model may be an offline training model. For example, the second neural network model may be a large-scale deep neural network model. Exemplarily, the second neural network model may be trained based on a large-scale unsupervised data set. The training process of the second neural network model may refer to the training process of the first neural network model above, and will not be repeated here.

[0066] According to the above description, in the embodiment of the present application, on the one hand, after obtaining the positioning parameter information, the positioning parameter information is first subjected to noise reduction processing based on the neural network model, and then the positioning parameter information after noise reduction processing is input into the positioning model for positioning, thereby helping to improve the accuracy of positioning. On the other hand, the neural network model can continuously learn and adjust the weights automatically through training to adapt to different usage scenarios and conditions, thereby improving the effect of noise reduction processing, thereby helping to improve the accuracy of positioning.

[0067] As mentioned above, the second neural network model is an offline training model. The scale of offline data information (or historical data information) used to train the offline training model will affect the prediction effect of the model. The positioning calibration in the related technology only relies on offline data information obtained in a single usage scenario (such as an experimental site) for model training. Since a single usage scenario cannot reflect the user's actual usage environment, the positioning model cannot effectively cover the user's multiple usage scenarios, resulting in a low accuracy of the positioning results predicted by the model.

[0068] In order to solve the above problems, in an embodiment of the present application, the offline data information used to train the second neural network model may include historical data information collected in multiple usage scenarios.

[0069] In some implementations, the second neural network model may be trained offline based on a third data set related to positioning, wherein the third data set includes offline data information measured in multiple usage scenarios.

[0070] The embodiments of the present application do not limit the specific content of the above-mentioned multiple usage scenarios. For example, multiple usage scenarios may include the surrounding environment of the experimental site, the surrounding environment of the user's home garage site, and the surrounding environment of the user's company parking lot. Multiple usage scenarios may include one or more factors that affect positioning accuracy mentioned above (such as multipath effect). In the above manner, the second neural network model can effectively cover the user's multiple usage scenarios, which helps to improve the accuracy of positioning.

[0071] Furthermore, in some implementations, the second neural network model can also be trained online based on the second data set. Online training of the second neural network model through online data information can enable the second neural network model to automatically learn and dynamically adjust weights. When the vehicle faces a new usage scenario, the second neural network model can be trained online based on the online data information of the usage scenario to optimize its own prediction performance to adapt to the factors affecting positioning accuracy in the new usage scenario, thereby helping to improve positioning accuracy.

[0072] The foregoing describes how the embodiments of the present application predict positioning results based on a neural network model, thereby obtaining first prediction information (such as coordinate information and / or positioning area information). The first prediction information can also be further optimized to improve positioning accuracy.

[0073] In some implementations, the first prediction information may be filtered based on a filtering model to obtain the second prediction information. The filtering model here may refer to a Kalman filter model. The classical Kalman filter has a better filtering effect on the signal, can effectively remove Gaussian noise and non-Gaussian noise, and is basically not affected by pulse signals, thereby helping to improve the accuracy of positioning.

[0074] For example, the forecast value of the positioning mode state at the current moment can be generated based on the positioning mode state at the previous moment. Then, the observation data is introduced, and the mode state is re-analyzed using the minimum variance estimation method. Then, as the mode state prediction continues and new observation data is input, the analysis process continues to evolve and the error is reduced to a minimum. Finally, the positioning coordinates and positioning area after the calculation are output. The positioning module state here can be a physical quantity that needs to be estimated, for example, it can include: coordinate information, speed, position direction, acceleration and other parameters of the device. The observation data here can be information obtained from various sensors, which is used to help the Kalman filter update its estimate of the positioning mode state. The observation data can include, for example: the rough position of the vehicle; information on acceleration and angular velocity, which can be used to estimate speed and direction; wheel speed information, which can be used to estimate the driving distance; information on the surrounding environment, which can be used to assist in positioning.

[0075] It should be understood that the embodiments of the present application do not limit the applicable wireless positioning technology. For example, the wireless positioning in the embodiments of the present application may be Bluetooth positioning or ultra wide band (UWB) positioning.

[0076] Combined with the following Figure 2 , a positioning method provided in another embodiment of the present application is introduced in detail with examples.

[0077] like Figure 2As shown, step S210, data collection. As the data source for algorithm processing, the following contents need to be completed:

[0078] (1) Summarize the RSSI information of Bluetooth anchor points: calculated by measuring the received wireless signal strength;

[0079] (2) Summarize the distance information of UWB anchor points: calculated based on the two-way time-of-flight method.

[0080] Step S220, data processing. There are two ways to perform noise reduction on the data. The first way includes online training and Gaussian filtering based on the first neural network model. The second way includes fusion of median filtering and Gaussian filtering. The specific processing methods involved in data processing are introduced below.

[0081] First, online training of the neural network model: The neural network model can continuously learn and adjust weights automatically through online training to optimize the ability of data noise reduction processing.

[0082] Fusion median: Fusion median filtering is a nonlinear smoothing technique that sets the grayscale value of each pixel to the median of all the pixel grayscale values ​​in a certain neighborhood window of the point. The basic principle of median filtering is to replace the value of a point in a digital image or digital sequence with the median of the values ​​of each point in a neighborhood of the point, so that the surrounding pixel values ​​are close to the true value, thereby eliminating isolated noise points.

[0083] Gaussian filtering: a linear smoothing filter, suitable for eliminating Gaussian noise, widely used in the noise reduction process of image processing. Gaussian filtering is the process of weighted averaging the entire image. The value of each pixel is obtained by weighted averaging the value of itself and other pixels in the neighborhood.

[0084] Step S230, establishing a fingerprint database. After the data processing is completed, a fingerprint database is established, and various wireless network structure features are used to explore advanced theories of information collection, processing and transmission to obtain the user's fine small-scale location information.

[0085] Step S240, offline training. Establish a second neural network model for offline training, decouple data collection and offline training, directly use the collected historical data for training, obtain a reinforcement learning (RL) strategy model with maximized positioning accuracy from the offline data, and perform repeated training to optimize the second neural network model.

[0086] Step S250, inferring positioning: After further tuning the neural network model through a combination of online training and offline training, the best positioning result is inferred.

[0087] Step S260: Dynamic optimization: After an accurate positioning result is inferred, it is necessary to dynamically update the positioning algorithm based on the Kalman filter model and the user's usage scenario to make the positioning algorithm more suitable for the user's usage.

[0088] Dynamic optimization is the process of predicting the true value based on the observed value and the estimated value. The classic Kalman filter has excellent signal filtering effect, relatively good real-time performance, minimal computational requirements, can effectively remove Gaussian noise and non-Gaussian noise, and is basically not affected by pulse signals.

[0089] Step S270, output the result. After the above process, the positioning coordinates and positioning area information are finally obtained.

[0090] Combined with the following Figure 3 , the implementation method of the positioning method provided in the embodiment of the present application is introduced in detail with examples.

[0091] like Figure 3 As shown, in step S310, the mobile phone calibration APP is connected to the vehicle anchor point via Bluetooth, and the coordinate data is sent to the anchor point in the form of broadcasting, and the anchor point parses the received RSSI data;

[0092] Step S320: If a self-calibration device is used, the coordinate data will be packaged and sent to a telematics service provider (TSP) in real time to save the coordinate data and perform time synchronization verification;

[0093] Data collection mainly includes outside and inside the car: outside the car, with the center of the car body as the center of the circle, assuming that the interval is 5 degrees and 0.5 meters, 360 / 5 = 72 points per circle, then a radius of 15 meters requires 15 / 0.5 × 72 = 2160 points, and the coordinates of each point are represented by a serial number. Each point collects data for typical scenarios such as facing, back, hands, pockets, and bags; inside the car, assuming 1.8 meters × 1.8 meters, 0.3 meters interval collection, the number of collected data at each point (classic scenario) is about 500.

[0094] Step S330, when the calibration data is uploaded through the mobile phone APP, the anchor point parses and processes the received RSSI data and sends it to the ternary content-addressable memory channel (TCAM) through the controller area network (CAN) interface.

[0095] In step S340, the calibration data uploaded by the mobile phone APP is packaged and processed through the TCAM network channel, and sent to the TSP platform regularly or in real time through NAD. The TSP platform performs structured data storage and provides interface capabilities in conjunction with front-end calls. After the calibration data collected by the self-calibration device is packaged, it is directly sent to the TSP for structured storage regularly or in real time. The TSP also provides an interface capability that can be queried by the front end.

[0096] Step S350: data collection is completed.

[0097] In step S360, the PC downloads data and the second neural network model for local training by calling the calibration interface of the TSP.

[0098] Step S370: The second neural network model training is completed and the positioning result is inferred.

[0099] Step S380, reverse loading the inferred positioning result into TSP to optimize the positioning accuracy.

[0100] In the related technologies, the positioning and calibration of vehicles is mainly done by manual measurement, and an average of four mobile phones can be calibrated per week, which requires a lot of manpower and time. At the same time, affected by the environment, operation, vehicle condition, etc., the calibrated data does not meet the expectations of market users.

[0101] The positioning method of the embodiment of the present application can quickly process the calibration algorithm and create the model, and use the neural network model for online training and offline training, so as to maximize the positioning accuracy and achieve the effect of improving the user experience.

[0102] At the same time, based on the rich model library created, iterative learning is carried out in combination with changes in user usage scenarios to improve user usage effects in different scenarios, greatly reduce calibration costs and improve efficiency.

[0103] Combination of the above Figures 1 to 3 , describes the method embodiment of the present application in detail, and the following is combined with Figures 4 to 5 , describes the device embodiment of the present application in detail. It should be understood that the description of the method embodiment corresponds to the description of the device embodiment, so the parts not described in detail can refer to the previous method embodiment.

[0104] Figure 4 This is a schematic diagram of the structure of the positioning device provided in the embodiment of the present application. Figure 4 As shown, the positioning device 400 includes an acquisition unit 410 , a first processing unit 420 and a second processing unit 430 .

[0105] An acquiring unit 410 is configured to acquire a first data set, where the first data set is online data information measured in a first usage scenario;

[0106] A first processing unit 420 is used to input the first data set into a first neural network model to obtain a second data set, wherein the first neural network model is used to perform noise reduction processing on the online data information;

[0107] The second processing unit 430 is used to input the second data set into a second neural network model to obtain first prediction information of the positioning result, and the second neural network model is an offline training model.

[0108] In some implementations, the first neural network model is trained online based on the first data set.

[0109] In some implementations, the second neural network model is trained offline based on a third data set, where the third data set is offline data information measured in multiple usage scenarios, and / or the second neural network model is trained online based on the second data set.

[0110] In some implementations, the determination unit 420 is further used to input the first data set into the first neural network model to obtain a fourth data set; and filter the first data values ​​in the fourth data set to obtain the second data set.

[0111] In some implementations, the first data value is RSSI data and / or distance information.

[0112] In some implementations, the method apparatus 400 is further configured to filter the first prediction information based on a filtering model to determine second prediction information.

[0113] In some implementations, the first data set includes RSSI data and coordinate information, and / or the first data set includes distance information and coordinate information.

[0114] Figure 5 It is a schematic diagram of the structure of a vehicle provided in an embodiment of the present application. Figure 5 The vehicle 500 shown may include: a memory 510, a processor 520, and an input / output interface 530. The memory 510, the processor 520, and the input / output interface 530 are connected through an internal connection path, the memory 510 is used to store instructions, and the processor 520 is used to execute the instructions stored in the memory 520 to implement the steps in the various methods described above. The processor 520 can also control the input / output interface 530 to receive input data and information, output data such as operation results, and control the transceiver to send signals.

[0115] It should be understood that in the embodiment of the present application, the processor 520 can adopt a general central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), or one or more integrated circuits to execute relevant programs to implement the technical solution provided in the embodiment of the present application.

[0116] The memory 510 may include a read-only memory and a random access memory, and provides instructions and data to the processor 520. A portion of the processor 520 may also include a nonvolatile random access memory. For example, the processor 520 may also store information on the device type.

[0117] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor 520 or an instruction in the form of software. The method for requesting uplink transmission resources disclosed in conjunction with the embodiment of the present application can be directly embodied as a hardware processor for execution, or a combination of hardware and software modules in the processor for execution. The software module can be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 510, and the processor 520 reads the information in the memory 510 and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it will not be described in detail here.

[0118] It should be understood that in the embodiments of the present application, the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0119] It should be understood that in the embodiment of the present application, "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0120] It should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0121] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0122] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0123] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0124] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that can be read by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0125] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A positioning method, characterized in that: include: Acquire a first data set, where the first data set is online data information measured in a first usage scenario; Inputting the first data set into a first neural network model to obtain a second data set, wherein the first neural network model is used to perform noise reduction processing on the online data information; The second data set is input into a second neural network model to obtain first prediction information of the positioning result, and the second neural network model is an offline training model.

2. The method according to claim 1, characterized in that The first neural network model is trained online based on the first data set.

3. The method according to claim 1 or 2, characterized in that: The second neural network model is trained offline based on a third data set, where the third data set is offline data information measured in multiple usage scenarios, and / or the second neural network model is trained online based on the second data set.

4. The method according to claim 1, characterized in that: The step of inputting the first data set into a first neural network model to obtain a second data set comprises: Inputting the first data set into the first neural network model to obtain a fourth data set; The first data value in the fourth data set is filtered to obtain the second data set.

5. The method according to claim 4, characterized in that The first data value is received signal strength indication RSSI data and / or distance information.

6. The method according to claim 1, characterized in that The method further comprises: The first prediction information is filtered based on a filtering model to obtain second prediction information.

7. The method according to claim 1, characterized in that: The first data set includes RSSI data and coordinate information; and / or, The first data set includes distance information and coordinate information.

8. A positioning device, characterized in that: include: An acquiring unit, configured to acquire a first data set, wherein the first data set is online data information measured in a first usage scenario; A first processing unit, configured to input the first data set into a first neural network model to obtain a second data set, wherein the first neural network model is configured to perform noise reduction processing on the online data information; The second processing unit is used to input the second data set into a second neural network model to obtain first prediction information of the positioning result, and the second neural network model is an offline training model.

9. A vehicle, characterized in that: The vehicle comprises a memory and a processor, the memory being used to store program codes, and the processor being used to call the program codes in the memory so that the vehicle executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium has a program code, and when the computer program code is executed on a computer, the computer is enabled to execute the method according to any one of claims 1 to 7.

Images