Positioning state determination method, apparatus and product

Abstract

This disclosure provides a method, apparatus, electronic device, storage medium, and computer program product for determining the positioning state, relating to the field of computer technology, specifically to intelligent driving, map navigation, intelligent transportation, large-scale modeling, and robotics, and applicable to positioning scenarios. The specific implementation scheme is as follows: A predictive environmental information of the target object in its predicted positioning state at the current moment is determined through an environmental prediction model; the uncertainty of the observation noise of the real environmental information is determined based on the constraint strength of the target object's real environmental information on the positioning estimation capability at the current moment; the predicted positioning state is adjusted based on the uncertainty of the observation noise and the difference between the predicted environmental information and the real environmental information, thus determining the final positioning state of the target object at the current moment. This disclosure improves the accuracy and precision of the positioning state and can effectively suppress the negative impact of environmental signal degradation areas on the final positioning state.

Description

Technical Field

[0001] This disclosure relates to the field of computer science, specifically to the fields of intelligent driving, map navigation, intelligent transportation, large-scale models, and robotics, and particularly to a method, device, electronic device, storage medium, and computer program product for determining positioning status, which can be applied to positioning scenarios. Background Technology

[0002] In some positioning systems, environmental signals (such as Wi-Fi, Bluetooth, and geomagnetic signals) can be incorporated into the state estimation framework as observations for positioning in GNSS (Global Navigation Satellite System) constrained environments. For example, environmental signals can be used as observations in filtering or optimization frameworks, with a fixed observation noise covariance matrix. However, in practical applications, the spatial distribution structure and information validity of environmental signals vary significantly across different environmental regions, directions, and times. A fixed observation noise covariance matrix approach cannot adapt to the characteristics of environmental signals changing with spatial location and time. In areas with insufficient information, this can introduce erroneous constraints on the positioning system, significantly reducing positioning accuracy. Summary of the Invention

[0003] This disclosure provides a method, apparatus, electronic device, storage medium, and computer program product for determining positioning status.

[0004] According to the first aspect, a method for determining the positioning state is provided, comprising: determining the predicted environmental information of the target object in the predicted positioning state at the current moment through an environmental prediction model, wherein the environmental prediction model is used to characterize the correspondence between the positioning state and the environmental information; determining the uncertainty of the observation noise of the real environmental information based on the constraint strength of the real environmental information of the target object on the positioning estimation capability at the current moment; adjusting the predicted positioning state based on the uncertainty of the observation noise and the difference between the predicted environmental information and the real environmental information, and determining the final positioning state of the target object at the current moment.

[0005] According to a second aspect, a device for determining a positioning state is provided, comprising: an environment prediction unit configured to determine predicted environmental information of a target object in a predicted positioning state at the current moment through an environment prediction model, wherein the environment prediction model is used to characterize the correspondence between the positioning state and the environmental information; an observation weighting unit configured to determine the uncertainty of the observation noise of the real environmental information based on the constraint strength of the target object's real environmental information on the positioning estimation capability at the current moment; and a positioning determination unit configured to adjust the predicted positioning state based on the uncertainty of the observation noise and the difference between the predicted environmental information and the real environmental information, and determine the final positioning state of the target object at the current moment.

[0006] According to a third aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a method as described in any implementation of the first aspect.

[0007] According to a fourth aspect, a non-transitory computer-readable storage medium is provided that stores computer instructions for causing a computer to perform the method described in any implementation of the first aspect.

[0008] According to a fifth aspect, a computer program product is provided, comprising: a computer program that, when executed by a processor, implements the method as described in any implementation of the first aspect.

[0009] According to the technology disclosed herein, a method and apparatus for determining a positioning state are provided. The method involves determining the predicted environmental information of a target object in its current predicted positioning state using an environmental prediction model, whereby the environmental prediction model characterizes the correspondence between the positioning state and the environmental information. Based on the constraint strength of the target object's actual environmental information on its positioning estimation capability at the current moment, the method determines the uncertainty of the observation noise of the actual environmental information. Based on the uncertainty of the observation noise and the difference between the predicted and actual environmental information, the method adjusts the predicted positioning state to determine the final positioning state of the target object at the current moment. By combining the environmental prediction model and the inversion positioning method, the final positioning state is output as a state estimate, exhibiting temporal continuity and improving the accuracy and precision of the positioning state. Furthermore, by dynamically determining the uncertainty of the observation noise of the actual environmental information based on the constraint strength of the actual environmental information on the positioning estimation capability, and adjusting the predicted positioning state accordingly to determine the final positioning state, the method effectively suppresses the negative impact of environmental signal degradation areas on the final positioning state, further improving the accuracy of the positioning state.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0011] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein: Figure 1 This is an exemplary system architecture diagram that can be applied to an embodiment of this disclosure; Figure 2 This is a flowchart of an embodiment of the method for determining the positioning status according to this disclosure; Figure 3 This is a schematic diagram illustrating an application scenario of the positioning status determination method according to this embodiment; Figure 4 This is a flowchart of yet another embodiment of the method for determining the positioning status according to this disclosure; Figure 5 This is a structural diagram of one embodiment of the positioning status determination device according to the present disclosure; Figure 6 This is a schematic diagram of the structure of a computer system suitable for implementing the embodiments of the present disclosure. Detailed Implementation

[0012] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0013] The technical solutions disclosed herein involve the collection, storage, use, processing, transmission, provision, and disclosure of various types of information, such as user personal information, in accordance with relevant laws and regulations and do not violate public order and good morals.

[0014] Figure 1 An exemplary architecture 100 is shown, which can be applied to the method and apparatus for determining the positioning status of the present disclosure.

[0015] like Figure 1As shown, the system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. The communication connections between terminal devices 101, 102, and 103 form a network topology. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0016] Terminal devices 101, 102, and 103 can be hardware or software that supports network connectivity for data acquisition, interaction, and processing. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices that support network connectivity, information acquisition, interaction, display, and processing functions, including but not limited to smartphones, tablets, e-book readers, laptops, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are imposed here.

[0017] Server 105 can be a server that provides various services, such as acquiring real-time environmental information sent by terminal devices 101, 102, and 103, and determining the final location status of the target object at the current moment based on the difference between real and predicted environmental information and the uncertainty of observation noise in real environmental information. As an example, server 105 can be a cloud server.

[0018] It should be noted that a server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules (such as software programs or software modules used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0019] It should also be noted that the method for determining the location status provided in the embodiments of this disclosure is generally executed by a server, but the possibility of it being executed by a terminal device, or by the server and the terminal device cooperating with each other, is not excluded. Accordingly, the various parts (e.g., various units) included in the location status determination device can be all set in the server, all set in the terminal device, or set in the server and the terminal device respectively.

[0020] It should be understood that Figure 1The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included. When the electronic device on which the location determination method runs does not need to transmit data with other electronic devices, the system architecture may only include the electronic device on which the location determination method runs (e.g., a server or terminal device).

[0021] Please refer to Figure 2 , Figure 2 A flowchart illustrating a method for determining a positioning state according to an embodiment of this disclosure. Flowchart 200 includes the following steps: Step 201: Determine the predicted environmental information of the target object in its current predicted positioning state using the environmental prediction model.

[0022] In this embodiment, the execution entity of the method for determining the positioning status (e.g., Figure 1 The server in the system determines the predicted environmental information of the target object in its current predicted location state through an environmental prediction model. The environmental prediction model characterizes the correspondence between the location state and the environmental information.

[0023] The target object is an object whose positioning status is to be determined in various positioning scenarios, such as intelligent terminals in unusable environments such as indoor buildings, underground shopping malls, subway stations or underground parking lots, mobile robots in warehouses, factories or large indoor venues, and vehicles in urban canyons or underground road scenarios.

[0024] This embodiment executes the discrete time-series iterative cycle. Each time the final positioning state at the current moment is determined, the final positioning state becomes the final positioning state at the previous moment of the next iteration cycle. This process is repeated cyclically to achieve continuous and uninterrupted positioning updates.

[0025] The time interval is the discrete sampling time. The sampling time interval between two adjacent times is a fixed or adaptive interval, which can be flexibly set according to the actual situation or equipment requirements, and is not limited here. Specifically, the previous time interval is the discrete sampling time before the current iteration cycle, and the current time interval is the discrete sampling time being calculated in the current iteration.

[0026] Predicted positioning state refers to the prior estimate of the pose information of the target object obtained solely through inertial recursion before the current environmental observation (real environmental information) is introduced for correction. It is the original prediction result based on the current real environmental data before the positioning update.

[0027] The final positioning state is the optimal and reliable posterior estimate of the target object after correction and optimization based on environmental observation information. It is the final positioning result output by the positioning system.

[0028] The positioning state (final positioning state and predicted positioning state) includes multiple state components, such as position component, attitude component, velocity component, and inertial zero bias component.

[0029] In a 3D scene, the positioning status is represented as follows:

[0030] in, For three-dimensional position; For three-dimensional velocity; The pose of a unit quaternion; Bias for the IMU (Inertial Measurement Unit) accelerometer; This is for biasing the IMU gyroscope. Typically... .

[0031] In a two-dimensional planar scene (such as indoor navigation), the positioning status is represented as:

[0032] in, For planar position, For heading angle, Linear velocity, ω is the angular velocity.

[0033] Predictive environmental information is based on an environmental prediction model and is derived directly from the predicted positioning state of the target object. It represents the environmental characteristic values ​​that should theoretically be observed at that location. It is used to characterize the inherent correspondence between a specific positioning state and environmental signals.

[0034] Environmental prediction models are mathematical models used to establish a stable mapping relationship between location status and environmental information, such as: Interpolation-based continuous field models construct global continuous interpolation functions based on spatial sampling points to achieve continuous prediction of environmental information under arbitrary positioning conditions.

[0035] The parameterized physical propagation model constructs an analytical expression based on the physical propagation laws of environmental signals, and directly calculates the corresponding environmental information from the positioning status.

[0036] The neural network model, based on coordinate encoding and multilayer perceptron to construct an implicit continuous mapping relationship, enables end-to-end prediction of environmental information under any positioning state.

[0037] Specifically, firstly, based on the target object's final positioning state at the previous moment and the inertial data of the target object between the previous moment and the current moment, the predicted positioning state of the target object at the current moment is determined; then, the predicted positioning state is input into the environmental prediction model, and the predicted environmental information of the target object in the predicted positioning state is determined through the environmental prediction model.

[0038] Inertial data is raw physical quantity data of motion acquired by inertial measurement units (such as triaxial accelerometers and triaxial gyroscopes) to characterize the changes in acceleration and angular velocity of a target object. It is the basic input for achieving inertial recursive predictive positioning. Inertial data includes, for example, triaxial acceleration data and triaxial angular velocity data.

[0039] For the process of determining the predicted positioning state: As an example, firstly, the final positioning state of the previous moment is used as the initial reference state for this prediction; then, inertial data such as continuous inertial acceleration and angular velocity output by the inertial measurement unit are collected during the entire time interval from the start of the previous moment to the end of the current moment; then, in chronological order, the inertial data within this time interval is integrated point by point in the time domain to sequentially calculate the velocity change, position change, and attitude rotation change of the target object within the time interval; finally, all the motion changes obtained by integration are sequentially superimposed onto the final positioning state of the previous moment to complete the inertial state recursion, thereby determining the predicted positioning state at the current moment.

[0040] As another example, using the final positioning state of the previous moment as the initial reference state, the entire segment of inertial data collected between the previous moment and the current moment is first preprocessed with smoothing filtering to remove random noise and high-frequency interference signals generated during inertial measurement, resulting in denoised smooth inertial data, ensuring the stability and accuracy of the inertial data. Then, based on the denoised smooth inertial data and combined with a preset inertial recursive model, the overall change parameters of the target object's motion attitude, velocity, and position from the previous moment to the current moment are calculated through a one-time overall calculation. This overall change parameter is then fused and adapted with the final positioning state of the previous moment, directly recursively obtaining the predicted positioning state of the current moment without point-by-point integration iteration, while effectively reducing the impact of noise on the predicted positioning accuracy.

[0041] For the environmental information prediction process based on environmental prediction models: As an example, during the offline phase, multiple sets of sample data containing known positioning states and corresponding environmental information are collected within the target scene. Using these samples as constraints, a global spatial interpolation continuous field is constructed as an environmental prediction model. This model can output continuous predicted environmental information values ​​for any location within the scene. During positioning, the predicted positioning state of the target object at the current moment is input into this spatial interpolation continuous field model. The model directly outputs the predicted environmental information corresponding to this predicted positioning state according to spatial interpolation rules, thus determining the predicted environmental information and providing a basis for subsequent positioning state adjustments.

[0042] As another example, based on the physical characteristics of environmental signals such as propagation attenuation and reflection in space, a parameterized physical propagation equation is constructed as an environmental prediction model. The equation uses the location state as the independent variable and environmental information as the dependent variable. During positioning, the predicted location state of the target object is substituted into this physical propagation equation, and the theoretical environmental signal strength, distribution characteristics, and other predicted environmental information at that location are calculated through the equation, realizing the rapid determination of predicted environmental information from the predicted location state.

[0043] In some optional implementations of this embodiment, the execution entity determines the predicted positioning state at the current moment in the following manner: First, based on the inertial data between the previous moment and the current moment, the positioning state increment of the target object at the current moment relative to the previous moment is determined; then, by combining the final positioning state of the target object at the previous moment and the positioning state increment, the predicted positioning state of the target object at the current moment is determined.

[0044] As an example, firstly, the inertial data between the previous moment and the current moment is pre-integrated to determine the positioning state increment of the target object at the current moment relative to the previous moment; then, based on the final positioning state of the target object at the previous moment, the positioning state increment is added to determine the predicted positioning state of the target object at the current moment.

[0045] As an example, the positioning state is predicted using the measurements from the inertial measurement unit (IMU) as process input. The IMU model is as follows:

[0046] in, , These are the raw measurements from the IMU. , For actual acceleration and angular velocity, Here is the attitude rotation matrix. The gravity vector , For bias, , To measure white noise.

[0047] At time step Internally, continuous integration is performed using IMU measurements (median value integration method):

[0048] in:

[0049] in, This is quaternion multiplication. This is the exponential mapping from a rotation vector to a rotation matrix (Rodrigues' formula). Indicates three-dimensional position, Represents three-dimensional velocity. Represents the attitude of a unit quaternion. Indicates IMU accelerometer bias. This indicates the IMU gyroscope bias.

[0050] In this implementation, the state increment is obtained by recursion based on inertial data, which enables continuous state prediction, ensuring uninterrupted positioning output in GNSS-constrained environments and improving positioning continuity and real-time performance.

[0051] In some optional implementations of this embodiment, the environmental prediction model adopts a Gaussian process regression model, which is obtained as follows: First, a sample set is obtained, which includes sample environmental data and sample location status. Then, the sample location status is input into a Gaussian process regression model to obtain environmental prediction data. Next, the probability distribution matching degree between the environmental prediction data and the sample environmental data corresponding to the sample location status is maximized to determine the standard deviation of signal amplitude, length scale, and observation noise. The probability distribution matching degree characterizes the consistency between the random distribution of the environmental prediction data and the actual distribution of the sample environmental data; the signal amplitude characterizes the overall fluctuation amplitude and intensity of change in the sample environmental data; and the length scale characterizes the spatial correlation distance of the sample environmental data. Finally, the Gaussian process regression model using the standard deviation of signal amplitude, length scale, and observation noise is determined as the environmental prediction model.

[0052] As an example, environmental signals are collected and a sample set is constructed using the following method.

[0053] In the target environment (indoors, underground parking lots, urban canyons, etc.), personnel or vehicles carrying environmental signal acquisition equipment traverse the scene along a planned path to collect environmental signal observation data. The acquisition equipment records the state reference value (sample positioning status, provided by high-precision GNSS, total station, or visual SLAM (Simultaneous Localization and Mapping)) and the corresponding environmental signal observation value (sample environmental data) for each acquisition point, forming a training sample set.

[0054] in, ( or ) is the first The reference positioning status (e.g., reference position) of each sampling point. For the sample environmental data at this location ( Dimensions, such as the RSSI (Received Signal Strength Indicator) of multiple APs (Access Points), and the three-axis components of the geomagnetic field.

[0055] Preprocessing of the collected data includes, but is not limited to: outlier removal, based on The three-standard-deviation criterion or the RANSAC (Random Sample Consensus) method is used to eliminate outlier observations; time alignment is performed to align signal observations with reference locations by timestamp; normalization is performed to standardize each signal component and eliminate dimensional differences.

[0056] Then, the environmental prediction model (continuous environmental field) is modeled in the following manner.

[0057] The first Each signal observation source (or signal component) is in the positioning state The true signal value at that location is modeled as a continuous field function. The actual observed signal is:

[0058] in, Zero-mean Gaussian observation noise. For all There are 1 signal component, in vector form:

[0059] in, , .

[0060] Then, for each signal component Independently establish a Gaussian process regression model:

[0061] in, It is the prior mean function, usually taken as , that is, the mean of the training set; For example, a kernel function can be used to describe the spatial correlation of observed signals, such as the Matérn-5 / 2 kernel.

[0062] in, express and The Euclidean distance between them This refers to the signal variance (hyperparameter). The length scale (hyperparameter) controls spatially relevant distances.

[0063] Given a training set In the new location status The posterior prediction at the location is:

[0064] The posterior mean and variance are as follows:

[0065] in, For sample environment data, For the kernel matrix ( ), This is the covariance vector between the new location and the training set.

[0066] The hyperparameters Optimization is achieved by maximizing the LML (Log Marginal Likelihood):

[0067] in, This is the position matrix of the sample set.

[0068] In this implementation, the hyperparameters of the model are adaptively optimized through Gaussian process regression to accurately fit the spatial environment distribution characteristics, thereby improving the accuracy of environmental prediction and enhancing the reliability of positioning environment modeling in complex scenarios.

[0069] In some optional implementations of this embodiment, the environmental prediction model employs a radial basis function network, and the environmental prediction model is obtained as follows: First, a sample set is obtained, in which the samples include sample environmental data and sample positioning status. Then, the sample positioning status is input into a radial basis function network to obtain environmental prediction data. Next, the error between the environmental prediction data and the sample environmental data corresponding to the sample positioning status is minimized to determine the weights of each basis function in the radial basis function network. Finally, the environmental prediction model is determined by combining each basis function and its weights.

[0070] In this implementation, the sample set and environmental prediction model can be built using the previous example, which will not be elaborated here.

[0071] Using a parameterized radial basis function network, the first... Each signal component is modeled as a weighted superposition of radial basis functions:

[0072] in, For the first Gaussian functions, Centered on (either uniformly distributed or determined by k-means clustering). For width parameter, The number of basis functions.

[0073] Model parameters Offline estimation using the least squares method:

[0074] in For designing a matrix That is, the result of substituting the a-th training point into the b-th basis function. This is the L2 regularization coefficient, used to prevent overfitting.

[0075] Radial basis function network models are computationally efficient and suitable for embedded scenarios with high real-time requirements; Gaussian process regression models offer better accuracy and provide uncertainty quantification, making them suitable for scenarios where accuracy is paramount.

[0076] In this implementation, a radial basis function network is used to fit the environmental distribution. The network weights are optimized by minimizing the error, resulting in fast modeling speed, strong real-time performance, and efficient and accurate prediction of environmental information.

[0077] In some examples, the aforementioned implementing entity can also implicitly model the environmental prediction model using a coordinate-based neural network:

[0078] in, The location or positioning state encoding function maps low-dimensional coordinates to a high-dimensional space, for example, as follows: For multilayer perceptrons, the SIREN (Sinusoidal Representation Networks) activation function is used. To better model high-frequency signal variations. Model parameters Training by minimizing prediction error:

[0079] in, Represented by regularization coefficient The regularization penalty term.

[0080] After obtaining the environmental prediction model through the above method, the forward observation model can be constructed in the following way:

[0081] in:

[0082] That is, the environmental prediction model, used to predict the predicted environmental information at the current predicted location state. dimension). The observation vector of the environmental signals acquired in real time. To observe noise.

[0083] The covariance matrix (uncertainty) of the observation noise is constructed as follows: For the radial basis function network model, a fixed empirical noise covariance is used. For Gaussian process regression models, the adaptive observation noise covariance can be further constructed using the posterior prediction variance:

[0084] in, This is the posterior variance of the Gaussian process regression model at the current predicted location, allowing the observation noise to automatically reflect the model's prediction confidence at that spatial location.

[0085] Step 202: Determine the uncertainty of the observation noise of the real environment information based on the constraint strength of the target object's real environment information on the positioning estimation capability at the current moment.

[0086] In this embodiment, the execution entity determines the predicted environmental information of the target object in the predicted positioning state through an environmental prediction model. The environmental prediction model is used to characterize the correspondence between the positioning state and the environmental information.

[0087] Real-world environmental information refers to environmental feature observation data directly collected by sensors when the target object is at its actual location at the current moment, providing objective observational basis for positioning calculation.

[0088] Positioning estimation capability refers to the ability of a positioning system to effectively solve, distinguish, and constrain the positioning state of a target object based on the real environmental information at the current moment. It is used to characterize the observability, solution accuracy, and stability of the positioning state based on the real environmental information at the current moment.

[0089] The constraint strength of real environmental information on positioning estimation capability refers to the strength, distinguishability, and contribution of the spatial distribution gradient and structural characteristics reflected by the environmental prediction model in the current predicted positioning state to the positioning state estimation. It is a quantitative indicator to measure whether environmental information can effectively constrain the positioning results.

[0090] Observation noise is a random measurement error that accompanies the observation of real environmental information; it is an unavoidable observation disturbance. The uncertainty of observation noise is a measure that quantifies the statistical distribution and error magnitude of observation noise, reflecting the reliability of the observed values. It can be directly determined by sensor calibration, statistical environmental measurements, or empirical parameters.

[0091] As an example, based on the observation sensitivity of the environmental prediction model in the current predicted positioning state, and combined with a preset basic observation noise uncertainty, a global constraint strength of environmental information on positioning estimation is constructed. Based on this global constraint strength, a monotonic mapping relationship is used to determine the observation noise uncertainty corresponding to the real environmental information, so that the observation noise uncertainty decreases as the constraint strength increases and increases as the constraint strength decreases, thereby achieving adaptive adjustment of the observation reliability. The observation sensitivity characterizes the sensitivity of the predicted environmental information to the predicted positioning state, and can be calculated based on the local rate of change of the environmental prediction model at the predicted positioning state, used to characterize the sensitivity of the predicted environmental information to changes in the positioning state.

[0092] As another example, based on the observation sensitivity of the environmental prediction model under the current predicted positioning state, the response strength of the real environmental information as the positioning state changes is determined, and this response strength is used as the constraint strength of the real environmental information on the positioning estimation capability. By using the joint mapping relationship between the constraint strength and the basic observation noise uncertainty, the final observation noise uncertainty is determined, so that the positioning system can improve the observation reliability when the constraint strength is high and reduce the observation reliability when the constraint strength is low, thus maintaining the stability of the positioning process.

[0093] In some optional implementations of this embodiment, the execution entity can perform step 202 to determine the uncertainty of the observation noise in the following manner: The first step is to determine the observation constraint matrix, which characterizes the strength of the constraint on the positioning estimation capability by real environmental information, based on the observation sensitivity and the basic uncertainty of the pre-set observation noise.

[0094] The observation constraint matrix is ​​a quantization matrix constructed by observation sensitivity and observation noise fundamental uncertainty. It is used to globally characterize the constraint capability, constraint distribution and constraint directionality characteristics formed by real environmental information on the positioning estimate in the current predicted positioning state, and reflects the comprehensive constraint effect of environmental information on each component of the positioning state.

[0095] The observation sensitivity is determined by the partial derivative of the environmental prediction model with respect to the current location state under the current predicted location state.

[0096] As an example, the observation constraint matrix is ​​the Fisher information matrix. Let the predicted positioning state at the current time be... The Fisher information matrix corresponding to a single environmental signal observation (real environmental information) is:

[0097] in, Indicates observational sensitivity. The fundamental uncertainty refers to the baseline uncertainty pre-set for the observation noise of the real environmental information without adaptive adjustment of the environmental observation information. It is the default covariance matrix of the observation noise and is used to represent the inherent measurement error level of the environmental sensor under standard operating conditions.

[0098] Describes the state At this point, the maximum amount of constraint information that a single environmental signal observation can provide for each state component is... The following is given: The Cramér-Rao lower bound (CRLB) achievable based on this observation, which is the theoretically achievable minimum positioning error:

[0099] in, Let be the covariance matrix of the positioning state estimate, representing the error distribution and uncertainty of the positioning state estimate result; It is a positive semi-definite partial order relation for matrices, indicating that the left matrix minus the right matrix equals a positive semi-definite matrix.

[0100] The second step is to decompose the observation constraint matrix to determine the maximum constraint strength that the real environment information can provide for each state component in the positioning estimation.

[0101] Maximum constraint strength refers to the upper limit of the constraint capability that environmental information can provide for positioning estimation in a certain state component or spatial direction of the positioning state. It is a measure of the maximum amount of effective information that environmental information in that direction can be used to constrain the positioning results.

[0102] The observation constraint matrix is ​​decomposed based on matrix numerical analysis to obtain the constraint strength distribution of each component of the corresponding positioning state. From this, the maximum constraint strength that real environmental information can provide for each state component in the positioning estimation is extracted.

[0103] As an example, for Eigenvalue decomposition is performed to reveal the information distribution in different directions:

[0104] in, Represents the set of directions, in which Indicates the first One direction; Representing the amount of information in each direction, it is a diagonal matrix with eigenvalues ​​on the diagonal. , , …that is, the maximum constraint strength in each direction.

[0105] Eigenvalues Reflecting real environmental information in corresponding feature directions ( The The amount of information provided in the column: The larger the value, the stronger the constraint that the real environment information in that direction has on the positioning status; This indicates that observability degrades in that direction, and real-world environmental information cannot provide effective constraints.

[0106] The third step is to determine the uncertainty of the observation noise based on the maximum constraint strength and the scene type in which the target object is located at the current moment.

[0107] The scenario types include isotropic scenarios, which represent real-world environmental information with consistent constraint strength in all directions, and anisotropic scenarios, which represent real-world environmental information with inconsistent constraint strength in all directions.

[0108] An isotropic scenario refers to a scenario where the constraint strength of environmental information is uniformly distributed in all directions of the positioning space, and the constraint capabilities on the positioning state components in each direction are similar. Examples include open halls, spacious squares, and large indoor public areas.

[0109] Anisotropic scenarios refer to scenarios where the constraint strength of environmental information varies significantly in different directions of the positioning space, resulting in an uneven constraint on the different directional components of the positioning state. Examples include narrow corridors, underground passages, urban canyon roads, and tunnels.

[0110] As an example, firstly, based on the maximum constraint strength corresponding to each state component, it is determined whether the current scene of the target object is an isotropic or anisotropic scene. If the numerical difference between the maximum constraint strengths corresponding to all state components is small, the target object is determined to be in an isotropic scene; if there is a significant numerical difference between the maximum constraint strengths corresponding to different state components, the target object is determined to be in an anisotropic scene.

[0111] Then, the uncertainty of the observation noise is determined as follows: In isotropic scenarios, the constraint strength of environmental information on each component of the positioning state is uniform overall. Therefore, a unified observation noise uncertainty matching the overall constraint strength is adopted. Based on the maximum constraint strength corresponding to all state components, the overall constraint level of the current environmental information is comprehensively determined, and the uncertainty of observation noise is uniformly configured according to this overall constraint level: the stronger the overall constraint strength, the smaller the uncertainty of observation noise; the weaker the overall constraint strength, the larger the uncertainty of observation noise, so as to ensure that the uncertainty of observation noise is consistent with the actual constraint capability provided by the environmental information.

[0112] In anisotropic scenarios, the constraint strength of environmental information on different components of the positioning state varies significantly. Therefore, an observation noise uncertainty adapted to the differences in constraint strength in each direction is adopted. Based on the maximum constraint strength corresponding to each state component, observation noise uncertainty components matching the constraint strength of each component are set: a smaller observation noise uncertainty is configured in the direction of the state component with higher constraint strength; and a larger observation noise uncertainty is configured in the direction of the state component with lower constraint strength. This allows the observation noise uncertainty to match the differentiated constraint strength in different directions, achieving an observation confidence configuration adapted to the constraint characteristics of environmental information.

[0113] In this implementation, the maximum constraint strength of each state component is obtained by constructing and decomposing the observation constraint matrix, and the observation noise uncertainty is determined by combining the differences in scene type. This can accurately match the environmental constraint characteristics and improve the rationality and stability of the positioning.

[0114] In some optional implementations of this embodiment, the execution entity can perform the third step described above to determine the uncertainty of the observation noise in the following manner: First, in response to the target object being in an isotropic scene at the current moment, the adjustment weight for the basic uncertainty is determined based on the maximum constraint strength; then, the uncertainty of the observation noise is determined by combining the adjustment weight and the basic uncertainty.

[0115] As an example, when the target object is in an isotropic scenario at the current moment, the adjustment weight for the basic uncertainty is determined by combining the maximum constraint strength corresponding to each state component; the basic uncertainty is scaled using this adjustment weight to obtain the observation noise uncertainty that matches the current constraint strength.

[0116] There is a negative correlation between the constraint strength of real environmental information and the uncertainty of observation noise; the stronger the constraint, the smaller the uncertainty of observation noise, and the weaker the constraint, the larger the uncertainty of observation noise.

[0117] In this implementation, the observation noise uncertainty is adaptively adjusted based on the maximum constraint strength in isotropic scenarios, which can match the observation weight with the environmental constraint capability, further improving the rationality of positioning updates and system stability.

[0118] In some optional implementations of this embodiment, the execution entity can determine the adjustment weight for the basic uncertainty in the following way: First, combine the maximum constraint strength corresponding to each state component to obtain the strength metric index; then, determine the adjustment weight based on the normalized strength metric index and the preset metric threshold.

[0119] The maximum constraint strength corresponding to each state component is multiplied or summed to obtain the strength metric index; the strength metric index is normalized, and the adjustment weight is determined based on the difference between the normalized strength metric index and the preset metric threshold.

[0120] As an example, firstly, we define several complementary intensity metrics: (1) D-optimality (overall information content): The "volume" of the Fisher information matrix reflects the combined effect of information in all directions; degradation occurs in any direction ( )hour, It is sensitive to weak directions.

[0121] (2) A-optimality (average information content): Summing the information content in each direction has strong robustness and is suitable as a comprehensive scoring index for the overall information content.

[0122] Then, we define a normalized information content score (a normalized intensity metric). :

[0123] or

[0124] in, The regularization coefficient (can be an empirical value or the same as the regularization coefficient) , A dimensionlessly consistent normalization constant is used to prevent zero denominators and compress the score to... The interval is used to facilitate subsequent weight calculation.

[0125] It should be noted that when the system is used simultaneously When using various environmental signals (such as simultaneous use of Wi-Fi and geomagnetic signals), the Fisher information matrix is ​​integrated. The sum of the information matrices of each signal source:

[0126] in, express The signal at the current moment The corresponding Fisher information matrix, express The signal at the current moment The corresponding observation sensitivity, express The signal at the current moment The corresponding fundamental uncertainty of observation noise.

[0127] The overall information content score is:

[0128] Then, define scalar weighting factors based on the normalized intensity metric. :

[0129] in: , is the steepness (slope) parameter, which controls the sensitivity of the weights to changes in information content; The information content threshold (preset measurement threshold) is set when... hour, It can be determined through offline statistics or cross-validation methods. and The value of .

[0130] The sigmoid function described above guarantees that the weights monotonically increase with the amount of information and remain numerically stable even in extreme cases. when ( When the value approaches 1 (indicating sufficient information and strong constraints), ; when ( When the value approaches 0 (indicating insufficient information and strong constraints), (Maintain minimum weight to avoid complete chain break).

[0131] Finally, based on the scalar weighting factor, an adaptive observation noise covariance matrix is ​​constructed:

[0132] Its physical meaning is: when At a higher level (sufficient information about the real environment and strong constraints), , The observations were fully utilized; when When the level is low (insufficient information about the real environment and weak constraint strength), , With the Kalman gain approaching zero, observation updates are effectively suppressed, and the positioning system relies primarily on prediction results.

[0133] In this implementation, the metric index is obtained by comprehensively considering the maximum constraint strength and the weight is calculated by normalization. This can stabilize and quantify the constraint level, making the adjustment of observation noise uncertainty more precise and reasonable, and improving the positioning robustness.

[0134] In some optional implementations of this embodiment, the execution entity can perform the third step described above to determine the uncertainty of the observation noise in the following manner: First, in response to the target object being in an anisotropic scenario at the current moment, a weight matrix representing the weight of each state component is determined based on the maximum constraint strength corresponding to each state component; then, the uncertainty of the observation noise is determined based on the weight matrix, the observation sensitivity, and the observation constraint matrix.

[0135] As an example, when the target object is in an anisotropic scene, a weight matching the constraint strength is assigned to each state component based on the maximum constraint strength corresponding to each component of the positioning state. The weights of each component are then combined according to the state dimension to form a weight matrix. This weight matrix is ​​used to weight and adjust the constraint distribution. Combining the observation sensitivity and the overall constraint characteristics of the observation constraint matrix, an observation noise uncertainty that can adapt to directional differences is calculated, so that weak constraint components correspond to higher uncertainty and strong constraint components correspond to lower uncertainty.

[0136] In this implementation, a weight matrix is ​​constructed according to components in anisotropic scenarios, which can differentiate the constraint strength in each direction, suppress the spread of errors in weak constraint directions, and make the positioning update more reasonable and the system more stable.

[0137] In some optional implementations of this embodiment, the execution entity can determine the weight matrix representing the weight of each state component in the following way: First, for each state component, determine the weight corresponding to the state component based on the maximum constraint strength of the state component; then, combine the weights of each state component to determine the weight matrix.

[0138] For each state component in the positioning state, a matching component weight is assigned based on its corresponding maximum constraint strength; the greater the constraint strength, the greater the component weight, and vice versa. The weights of all state components are arranged in order according to their corresponding positions and combined to form a weight matrix consistent with the positioning state dimension, which is used to characterize the difference in constraint contribution of each state component.

[0139] As an example, define a directional weight matrix. (Positive definite symmetry):

[0140] in, Represents the set of directions. Scalar weighting functions in each direction Defined as:

[0141] in, Regularization parameters (dimensions and dimensions) Same). When hour, (Make full use of this directional constraint); when hour, (Suppress weak constraints in this direction).

[0142] Uncertainty of adaptive equivalent observation noise in the corresponding direction:

[0143] in, Indicates observational sensitivity.

[0144] Or, equivalently, in the EKF (Extended Kalman Filter) information filtering form, the adaptive information matrix is ​​directly defined as:

[0145] in, This represents the Hadamard product (element-by-element multiplication). Used to apply differentiated weights to information matrix components in different directions.

[0146] In this implementation, weights are assigned separately to state components and a weight matrix is ​​constructed, which can accurately reflect the differences in constraints of each component and provide a reliable basis for adjusting the differentiated observation noise in anisotropic scenarios.

[0147] In some optional implementations of this embodiment, the execution entity can determine the uncertainty of observation noise based on the weight matrix, observation sensitivity, and observation constraint matrix in the following way: First, adjust the maximum constraint strength corresponding to each state component according to the level of the maximum constraint strength corresponding to each state component to obtain the adjusted observation constraint matrix of the built-in weight matrix; then, determine the uncertainty of observation noise based on the observation sensitivity and the adjusted observation constraint matrix.

[0148] As an example, the maximum constraint strength corresponding to each positioning state component is divided into corresponding strength levels. The original maximum constraint strength of each component is then adjusted differentially according to different strength levels. During the adjustment process, the weight matrix corresponding to each state component is embedded and fused into the constraint strength, completing the optimization and reconstruction of the original observation constraint matrix. This yields an adjusted observation constraint matrix with an embedded weight matrix that reflects the differences in constraint characteristics across directions. Finally, by combining the known observation sensitivity with the adjusted observation constraint matrix, a joint operation is performed to determine the observation noise uncertainty that can adapt to the directional constraint characteristics of anisotropic scenes.

[0149] In this implementation, the constraint strength is adjusted according to the constraint level and the observation constraint matrix is ​​reconstructed by embedding a weight matrix. This can finely adapt to the directional differences in anisotropic scenarios and improve the accuracy of determining the uncertainty of observation noise and the stability of positioning.

[0150] In some optional implementations of this embodiment, the execution entity can determine the adjusted observation constraint matrix of the built-in weight matrix in the following way: First, based on the comparison results of the preset strength threshold and the maximum constraint strength corresponding to each state component, each state component is divided into at least one component group; then, the maximum constraint strength corresponding to the state component in the component group is adjusted by adopting the adjustment method corresponding to each component group, and the adjusted constraint strength of the weight corresponding to the built-in state component is determined; finally, the adjusted observation constraint matrix is ​​obtained by combining the adjusted constraint strength corresponding to each state component.

[0151] As an example, the maximum constraint strength of each state component is compared with a preset strength threshold one by one. Based on the numerical magnitude, all state components are divided into different component groups, with state components within the same component group having similar constraint strength levels. For each component group, a corresponding differentiated adjustment rule is set, and the original maximum constraint strength of each state component within the group is adjusted according to the adjustment method of its respective component group, resulting in an optimized adjusted constraint strength with embedded corresponding component weights. Finally, the adjusted constraint strengths of all state components are reconstructed and arranged according to their original dimensional order to generate a complete adjusted observation constraint matrix.

[0152] The number of component groups can be flexibly set according to the actual situation and is not limited here. When the number of component groups is greater than 2, it is necessary to divide the numerical range of each component group into multiple preset intensity thresholds.

[0153] As an example, the preset intensity threshold is For maximum constraint strength Categorize: Corresponding feature direction This represents the direction of observable degradation.

[0154] Taking a corridor scene as an example, directional imbalance can be modeled as follows: Lateral (perpendicular corridor direction): Large environmental signal gradient, large amount of information. Longitudinal (along the corridor): Small environmental signal gradient, small amount of information. Information distribution ellipse: It exhibits obvious directional stretching.

[0155] Soft suppression of information in the degradation direction is applied to construct an adjusted observation constraint matrix. :

[0156] This correction method maintains the original constraints for the strong information direction and adjusts them according to the weak information direction. The squared decay suppression is performed proportionally, which ensures the positive definiteness and numerical stability of the correction matrix.

[0157] The weights corresponding to the built-in state components refer to the soft suppression coefficients of the degenerate components. The adaptive weights corresponding to each state component are embedded into the adjusted constraint strength; the weights for non-degenerate strong constraint directions are approximately 1, while the weights for degenerate weak constraint directions are proportionally reduced.

[0158] The uncertainty of the equivalent observation noise is derived from the adjusted observation constraint matrix and used to replace the uncertainty in the standard EKF update. :

[0159] Will By substituting into the EKF update equation, differentiated information equilibrium can be achieved in each direction.

[0160] In this implementation, by adjusting the constraint strength of each component through threshold grouping and hierarchical adjustment, fine-grained hierarchical weight adaptation can be achieved, the characteristics of the observation constraint matrix can be optimized, and the accuracy and robustness of observation noise determination in anisotropic scenarios can be further improved.

[0161] Step 203: Based on the uncertainty of the observation noise and the difference between the predicted environmental information and the actual environmental information, adjust the predicted positioning state and determine the final positioning state of the target object at the current moment.

[0162] In this embodiment, the execution entity adjusts the predicted positioning state based on the difference between the predicted environmental information and the actual environmental information of the target object at the current moment, and determines the final positioning state of the target object at the current moment.

[0163] The difference is the numerical deviation between predicted environmental information and actual environmental information in the same dimension, used to quantify the degree of deviation between predicted environmental values ​​and actual observed values.

[0164] For example, the difference at the current time can be calculated using the following formula. :

[0165] in, This represents the current real-time environmental information. Environmental prediction model Predicted positioning status at the current moment The following is the predicted environmental information.

[0166] As an example, firstly, the differences between the predicted and actual environmental information across various environmental feature dimensions are calculated to form a one-dimensional or multi-dimensional residual vector. Then, this residual vector and the uncertainty vector of the observation noise are used as the basis for positioning state correction. Following a preset linear correction rule, the predicted positioning state is adjusted component-by-component along the spatial direction pointed to by the residuals. Finally, after component-by-component correction, the corrected positioning state is verified for reasonableness. If the verification passes, it is determined as the final positioning state of the target object at the current moment.

[0167] As another example, starting from the predicted positioning state at the current moment, a local search region is established within the positioning feasible space covered by the environmental prediction model. Multiple candidate positioning states are traversed within this local search region. For each candidate positioning state, the corresponding candidate predicted environmental information is calculated using the environmental prediction model, and the difference between the candidate predicted environmental information and the actual environmental information is determined. The difference is weighted by incorporating the uncertainty of observation noise to obtain the weighted deviation degree corresponding to each candidate positioning state. The candidate positioning state with the smallest weighted deviation degree is selected as the optimal matching position, and this optimal matching position is determined as the final positioning state of the target object at the current moment.

[0168] In some optional implementations of this embodiment, before executing step 203, the execution entity further performs the following operation: determining the uncertainty of the predicted positioning state based on the uncertainty of the final positioning state of the target object at the previous moment, the uncertainty of the state transition function, and the uncertainty of noise in the inertial data acquisition process, wherein the state transition function characterizes the correspondence between the final positioning state at the previous moment, the inertial data, and the predicted positioning state.

[0169] The uncertainty of the final positioning state at the previous moment is used to quantify the reliability of the target object's final positioning state at the previous moment and the range of error distribution. It is output from the positioning solution process at the previous moment and can be directly inherited and used.

[0170] The state transition function is a function that represents the mapping relationship between the final positioning state at the previous moment, the inertial data between the previous moment and the current moment, and the predicted positioning state at the current moment. It can be pre-constructed based on the laws of inertial motion and the positioning state.

[0171] Noise in the inertial data acquisition process is the random measurement error that occurs when the inertial sensor acquires inertial data. It is generally determined by the inherent characteristics of the inertial sensor and the working environment.

[0172] The uncertainty of noise during inertial data acquisition is a measure used to quantify the fluctuation amplitude and statistical distribution of inertial measurement noise, which can be obtained based on sensor parameter calibration results or experimental statistics.

[0173] As an example, firstly, the uncertainty of the target object's final positioning state at the previous moment, the pre-constructed state transition function, and the uncertainty of noise during inertial data acquisition are obtained. Then, based on the uncertainty propagation characteristics of the state transition function, the uncertainty of the final positioning state at the previous moment is statistically propagated through the state transition process. Simultaneously, the uncertainty of noise during inertial data acquisition is incorporated into the propagation calculation according to its impact on positioning state prediction. Finally, by comprehensively superimposing and propagating these two types of uncertainties during the state prediction process, the uncertainty of the target object's predicted positioning state at the current moment is obtained.

[0174] In this implementation, the execution entity performs step 203 as follows to determine the final positioning status of the target object at the current moment: based on the uncertainty and difference of the observation noise of the real environmental information, the uncertainty of the predicted positioning status and the observation sensitivity, the predicted positioning status is adjusted to determine the final positioning status of the target object at the current moment.

[0175] Observation sensitivity characterizes the sensitivity of predicted environmental information to predicted positioning status. It can be calculated based on the local rate of change of the environmental prediction model at the predicted positioning status and is used to characterize the sensitivity of predicted environmental information to changes in positioning status.

[0176] Referring to the above environmental prediction model, the observation sensitivity can be represented by its Jacobian matrix:

[0177] Wherein, the position Jacobian matrix is , its first Behavior No. The partial derivatives of each signal component with respect to position.

[0178] For the radial basis function network model, the position Jacobian matrix is:

[0179] in, Indicates the first Predictive environmental information for each signal component This indicates the predicted positioning status at the current moment. Indicates the number of radial basis functions. Indicates the first The signal component, the first The weights of the basis functions, Indicates the first The signal component, the first The center of each basis function Indicates the first The signal component, the first The width scale (length scale) of each basis function.

[0180] For the Gaussian process regression model, the location Jacobian matrix is:

[0181] in, This represents the covariance vector between the predicted positioning state and the positioning state of each sample. Indicates the first The covariance matrix of the sample set under each signal component; Indicates the first Standard deviation of observation noise for each signal component; Represents the identity matrix; Indicates the first A vector of sample environmental observation values ​​for each signal component.

[0182] Matérn-5 / 2 checks the partial derivative of the positioning state, letting ,but:

[0183] in, This represents the Matérn-5 / 2 covariance kernel function of the Gaussian process regression model. This indicates that the Matérn-5 / 2 covariance kernel function is effective for predicting the localization state. The partial derivatives are used to solve for the observation sensitivity. Indicates the first The sample localization state corresponding to each training sample. This represents the Euclidean distance between the predicted positioning state and the sample positioning state. Indicates the first The signal amplitude of each signal component Indicates the first The length scale of each signal component.

[0184] Observation noise is a random measurement error that accompanies the observation of real environmental information; it is an unavoidable observation disturbance. The uncertainty of observation noise is a measure that quantifies the statistical distribution and error magnitude of observation noise, reflecting the reliability of the observed values. It can be directly determined by sensor calibration, statistical environmental measurements, or empirical parameters.

[0185] As an example, firstly, the differences between the predicted and actual environmental information, the observation noise uncertainty of the actual environmental information, the uncertainty of the predicted positioning state, and the observation sensitivity are obtained. Then, based on the observation sensitivity, the uncertainty of the predicted positioning state, and the observation noise uncertainty, the correction strength for the difference on the positioning state is determined. Finally, based on this correction strength and the difference value, the predicted positioning state is adaptively adjusted to obtain the final positioning state of the target object at the current moment.

[0186] In this implementation, by introducing uncertainty in observation noise, uncertainty in predicted positioning state, and observation sensitivity for state adjustment, the reliability and continuity of positioning results are improved, the robustness of the system is enhanced, and the positioning output is made more stable and more interpretable.

[0187] In some optional implementations of this embodiment, the execution entity determines the uncertainty of the predicted positioning state in the following manner: First, based on the state transition function, the state transition sensitivity and noise input sensitivity are determined. State transition sensitivity characterizes how sensitive the predicted positioning state is to the final positioning state at the previous time step. It is obtained by calculating the rate of change of the state at the final positioning state at the previous time step, based on the state transition function.

[0188] Noise input sensitivity characterizes the sensitivity of the predicted positioning state to noise during inertial data acquisition. It is calculated by taking the partial derivative of the inertial data noise term with respect to the state transition function.

[0189] Then, based on the uncertainty of the final positioning state at the previous moment, the state transition sensitivity, the noise input sensitivity, and the uncertainty of noise during the inertial data acquisition process, the uncertainty of the predicted positioning state is determined.

[0190] Specifically, the uncertainty of the target object's final positioning state at the previous moment, the pre-constructed state transition function, and the uncertainty of noise during inertial data acquisition are obtained. Based on the state transition function, the state transition sensitivity and noise input sensitivity are calculated at the current recursive position. The uncertainty of the final positioning state at the previous moment is propagated through the state transition sensitivity; simultaneously, the uncertainty of noise during inertial data acquisition is introduced through the noise input sensitivity. The above-mentioned propagated and introduced uncertainties are then fused and statistically calculated to finally obtain the uncertainty of the target object's predicted positioning state at the current moment.

[0191] The state transition process is represented as follows:

[0192] in, For the current moment Predicted positioning status, For the previous moment The final positioning status; This is the state transition function; The control input vector for the inertial measurement unit (IMU) is a combination of the IMU's original measurement values. These represent the triaxial acceleration measurement (transposed) of the accelerometer and the triaxial angular velocity measurement (transposed) of the gyroscope, respectively.

[0193] The uncertainty of the predicted positioning state is calculated using the following formula:

[0194] in, Let be the prediction state error covariance matrix, representing the uncertainty of the predicted positioning state at the current moment. Let be the state transition Jacobian matrix, representing the state transition sensitivity; The noise input Jacobian matrix represents the noise input sensitivity. The process noise covariance includes the measurement noise of the inertial measurement unit and the bias random walk noise, representing the uncertainty of noise during inertial data acquisition. This is the covariance matrix for the predicted state error.

[0195] In this implementation, uncertainty is accurately propagated through state transition sensitivity and noise input sensitivity, reliably quantifies the range of predicted state error, improves prediction reliability, and provides stable support for subsequent positioning correction.

[0196] In some optional implementations of this embodiment, the execution entity determines the final positioning state of the target object at the current moment in the following manner: First step, determine the uncertainty of the difference based on the uncertainty of the observation noise, the uncertainty of the predicted positioning state, and the observation sensitivity; Second step, determine the correction weight of the real environment information relative to the predicted positioning state by combining the uncertainty of the difference, the observation sensitivity, and the uncertainty of the predicted positioning state; Third step, determine the final positioning state at the current moment based on the predicted positioning state, the correction weight, and the difference.

[0197] The uncertainty of the difference is used to quantify the credibility and error distribution range of the difference between the predicted environmental information and the actual environmental information, and reflects the statistical stability of the difference value.

[0198] Specifically, firstly, based on the uncertainty of observation noise, the uncertainty of predicted positioning state, and observation sensitivity, statistical fusion calculations are performed to obtain the uncertainty of the difference between predicted and actual environmental information. Next, based on this uncertainty, observation sensitivity, and the uncertainty of predicted positioning state, a comprehensive weighted calculation is performed to determine the correction weight of actual environmental information on the predicted positioning state. Finally, based on the predicted positioning state, and combining the correction weights and the difference value, state correction is performed to obtain the final positioning state of the target object at the current moment.

[0199] As an example, first, the uncertainty of the difference is calculated using the following formula:

[0200] in, Indicates the uncertainty of the difference. This represents the uncertainty of the predicted positioning state at the current moment. This represents the uncertainty of the observation noise.

[0201] Then, the Kalman gain (corrected weights) is calculated using the following formula. :

[0202] Finally, the state estimate is updated using the following formula:

[0203] in, This indicates the final location status at the current moment. This indicates the predicted positioning status at the current moment.

[0204] In this implementation, the correction weight is adaptively determined based on uncertainty and observation sensitivity to achieve accurate state updates, improve positioning continuity and reliability, and enhance the system's robustness to environmental disturbances.

[0205] In some optional implementations of this embodiment, the execution entity further performs the following operation: determining the uncertainty of the final positioning state at the current moment based on the uncertainty of the predicted positioning state at the current moment, the correction weight, the observation sensitivity, and the uncertainty of the observation noise.

[0206] Using observation sensitivity as the transmission constraint between positioning status and environmental observation, and using the correction weight as the fusion coefficient, the uncertainty of predicted positioning status and the uncertainty of observation noise are statistically propagated and fused to obtain the uncertainty of the final positioning status at the current moment.

[0207] As an example, the uncertainty of the final positioning state at the current moment is calculated using the following formula. :

[0208] in, This is the normalization matrix used for aligning the dimensions of the state vector.

[0209] This implementation method achieves precise quantification of the uncertainty of the final positioning state at each moment, continuously outputs credibility indicators, improves the interpretability of positioning results and system reliability, and can support subsequent stable decision-making.

[0210] In some optional implementations of this embodiment, the execution entity also performs the following operation: determining the validity of the real environment information based on the difference and the uncertainty of the difference.

[0211] Specifically, first, the difference between the predicted environmental information and the actual environmental information is determined, and a validity assessment index is constructed based on the uncertainty of the difference. Then, this validity assessment index is compared with a preset judgment threshold. If the validity assessment index is less than or equal to the threshold, the actual environmental information is determined to be a valid observation; if it is greater than the threshold, the actual environmental information is determined to be an invalid observation, thus completing the determination of the validity of the actual environmental information.

[0212] In this implementation, the aforementioned execution entity performs step 203 as follows to determine the final location status of the target object at the current moment: in response to the effective observation of real environmental information, the predicted location status is adjusted according to the difference to determine the final location status of the target object at the current moment.

[0213] As an example, the following formula is used to calculate the Mahalanobis distance test for real-world environmental information to determine the validity assessment index. :

[0214] like (judgment threshold, Distributed in degrees of freedom Significance level The critical value below, usually If the observation (real environmental information) is considered an anomalous observation (possibly caused by environmental signal fluctuations, multipath effects, or model failure), the final positioning status will not be determined based on the difference; only the predicted positioning status will be retained, thereby avoiding outlier contamination of the positioning status.

[0215] like If the observation is valid, the predicted positioning status is adjusted based on the difference to determine the final positioning status of the target object at the current moment.

[0216] It should be noted that this implementation method can be combined with the above implementation methods. For example, when the real environmental information is determined to be a valid observation, the predicted positioning state can be adjusted according to the difference, the uncertainty of the observation noise of the real environmental information, the uncertainty of the predicted positioning state, and the observation sensitivity to determine the final positioning state of the target object at the current moment.

[0217] This implementation improves system robustness by validating real-world environmental information, eliminating interference from abnormal observations, avoiding outliers from contaminating positioning results, and ensuring continuous, stable, and reliable positioning.

[0218] When GNSS signals are available, such as in an unobstructed outdoor environment or an environment that is partially obstructed but still receivable, GNSS-related observations are introduced into the positioning system as additional observation information. This information is then fused with the original observation information within the same EKF framework to achieve seamless indoor and outdoor positioning and ensure the continuity of positioning.

[0219] Specifically, when GNSS signals are available (such as outdoors or in partially obscured environments), GNSS pseudorange / carrier phase or GNSS calculated position is introduced as additional observation:

[0220] The corresponding Jacobi is It directly performs observation fusion within the same positioning optimization framework to achieve seamless indoor and outdoor positioning. This represents a 3rd order identity matrix.

[0221] in, These are GNSS observations. This is the GNSS observation matrix. Positioning status, GNSS observation noise

[0222] In this embodiment, to further improve the stability of the positioning system, especially the positioning reliability when GNSS signals are missing or environmental observation signals are interfered with, observation information from vehicle motion models or wheel speedometers is introduced to constrain the speed state in the positioning system. Similarly, multi-source information fusion is achieved within the EKF framework.

[0223] Introduce a vehicle motion model or wheel speedometer to observe the constrained speed state:

[0224] in, Represents the observed velocity value. Indicates speed status. Indicates velocity observation noise, This represents the vehicle velocity observation mapping matrix. It is derived from the attitude rotation matrix. The vehicle velocity is converted to the reference coordinate system and then fused within the EKF framework.

[0225] In this embodiment, to meet the requirements of high-precision positioning, a factor graph optimization framework can be introduced to replace the single EKF fusion framework, thus adapting to higher-precision positioning scenarios.

[0226] In addition to EKF, the forward observation model in this embodiment can also incorporate factor graph optimization frameworks, such as GTSAM (an open-source map building library) and g2o (General Graph Optimization), for batch optimization or sliding window optimization, which is suitable for scenarios with higher accuracy requirements or those requiring smooth historical trajectories.

[0227] In the factor diagram, each environmental signal observation constitutes an environmental field observation factor, and its residual is:

[0228] The information matrix is It participates in joint optimization along with other factors (IMU factor, odometry factor, map constraint factor, etc.):

[0229] in, Let be the set of positioning states to be optimized. The optimal positioning state is obtained through joint optimization. This represents finding the optimal state that minimizes the overall cost function. Indicates the first A continuous field residual, Indicates the first IMU factor residuals.

[0230] See also Figure 3 , Figure 3This is a schematic diagram of an application scenario 300 of the positioning state determination method according to this embodiment. A user drives vehicle 301 in a GNSS-constrained environment. For each moment, server 302 determines the predicted positioning state of vehicle 301 at the current moment based on the final positioning state of vehicle 301 at the previous moment and the inertial data of vehicle 301 between the previous and current moments; it determines the predicted environmental information of the target object in the predicted positioning state through an environmental prediction model, where the environmental prediction model characterizes the correspondence between the positioning state and the environmental information; it determines the uncertainty of the observation noise of the real environmental information based on the constraint strength of the target object's real environmental information on the positioning estimation capability, where the constraint strength is negatively correlated with the uncertainty of the observation noise; it adjusts the predicted positioning state based on the difference between the predicted environmental information and the real environmental information collected by vehicle 301 at the current moment, and determines the final positioning state of vehicle 301 at the current moment. The navigation application in vehicle 301 performs navigation based on the final positioning state at each moment.

[0231] In this embodiment, the predicted environmental information of the target object in its predicted positioning state at the current moment is determined by an environmental prediction model. The environmental prediction model characterizes the correspondence between the positioning state and the environmental information. Based on the constraint strength of the target object's actual environmental information on the positioning estimation capability at the current moment, the uncertainty of the observation noise of the actual environmental information is determined. Based on the uncertainty of the observation noise and the difference between the predicted environmental information and the actual environmental information, the predicted positioning state is adjusted to determine the final positioning state of the target object at the current moment. Thus, by combining the environmental prediction model and the inversion positioning method, the final positioning state is output as a state estimate, exhibiting temporal continuity and improving the accuracy and precision of the positioning state. Furthermore, by dynamically determining the uncertainty of the observation noise of the actual environmental information based on the constraint strength of the actual environmental information on the positioning estimation capability, and adjusting the predicted positioning state accordingly to determine the final positioning state, the negative impact of environmental signal degradation areas on the final positioning state can be effectively suppressed, further improving the accuracy of the positioning state.

[0232] Continue to refer to Figure 4 The illustration shows a schematic flow 400 of yet another embodiment of the method for determining the positioning status according to the present disclosure. Flow 400 includes the following steps: Step 401: Based on the inertial data of the target object between the previous moment and the current moment, determine the positioning state increment of the target object at the current moment relative to the previous moment.

[0233] Step 402: Combine the target object's final location status and location status increment at the previous moment to determine the target object's predicted location status at the current moment.

[0234] Step 403: Determine the predicted environmental information of the target object in the predicted positioning state through the environmental prediction model.

[0235] The predictive environmental information of the target object in the predicted positioning state is determined by the environmental prediction model.

[0236] Step 404: Based on the observation sensitivity and the basic uncertainty of the observation noise, determine the observation constraint matrix that characterizes the constraint strength of the real environment information on the positioning estimation capability.

[0237] Among them, observation sensitivity characterizes the sensitivity of the predicted environmental information to the predicted positioning status.

[0238] Step 405: Decompose the observation constraint matrix to determine the maximum constraint strength that the real environment information can provide for each state component in the positioning estimation.

[0239] Step 406: Determine the uncertainty of the observation noise based on the maximum constraint strength and the scene type in which the target object is located at the current moment.

[0240] The scenario types include isotropic scenarios, which represent real-world environmental information with consistent constraint strength in all directions, and anisotropic scenarios, which represent real-world environmental information with inconsistent constraint strength in all directions.

[0241] Step 407: Determine the state transition sensitivity and noise input sensitivity based on the state transition function.

[0242] The state transition function characterizes the correspondence between the final positioning state, inertial data, and predicted positioning state at the previous moment. The state transition sensitivity characterizes the sensitivity of the predicted positioning state to the final positioning state at the previous moment. The noise input sensitivity characterizes the sensitivity of the predicted positioning state to noise during the inertial data acquisition process.

[0243] Step 408: Determine the uncertainty of the predicted positioning state based on the uncertainty of the final positioning state at the previous moment, the state transition sensitivity, the noise input sensitivity, and the uncertainty of noise during the inertial data acquisition process.

[0244] Step 409: Determine the uncertainty of the difference between the predicted environmental information and the actual environmental information of the target object at the current moment, based on the uncertainty of the observation noise, the uncertainty of the predicted positioning state, and the observation sensitivity.

[0245] Step 410: Combine the uncertainty of the difference, the observation sensitivity, and the uncertainty of the predicted positioning state to determine the correction weight of the real environment information relative to the predicted positioning state.

[0246] Step 411: Determine the final positioning status at the current moment based on the predicted positioning status, corrected weights, and differences.

[0247] Step 412: Determine the uncertainty of the final positioning state at the current moment based on the uncertainty of the predicted positioning state at the current moment, the correction weight, the observation sensitivity, and the uncertainty of the observation noise.

[0248] The aforementioned execution entity can iteratively execute steps 401-412 to determine the final positioning status at each moment.

[0249] The process 400 of the positioning state determination method in this embodiment, compared with the above-described process 200, specifically describes the uncertainty of observation noise, the process of determining the predicted positioning state, and the process of determining the final positioning state. Thus, based on the constraint strength of the real environmental information on the positioning estimation capability, the uncertainty of the observation noise of the real environmental information is dynamically determined, and the predicted positioning state is adjusted accordingly to determine the final positioning state. This can effectively suppress the negative impact of environmental signal degradation areas on the final positioning state and further improve the accuracy of the positioning state.

[0250] Continue to refer to Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides an embodiment of a positioning state determination device, which is similar to... Figure 2 Corresponding to the method embodiments shown, the system can be specifically applied to various electronic devices.

[0251] like Figure 5 As shown, the positioning state determination device 500 includes: an environment prediction unit 501, configured to determine the predicted environment information of the target object in the predicted positioning state at the current moment through an environment prediction model, wherein the environment prediction model is used to characterize the correspondence between the positioning state and the environment information; an observation weighting unit 502, configured to determine the uncertainty of the observation noise of the real environment information based on the constraint strength of the target object's real environment information on the positioning estimation capability at the current moment; and a positioning determination unit 503, configured to adjust the predicted positioning state based on the uncertainty of the observation noise and the difference between the predicted environment information and the real environment information, and determine the final positioning state of the target object at the current moment.

[0252] In some optional implementations of this embodiment, the observation weighting unit 502 is further configured to: determine an observation constraint matrix characterizing the constraint strength of the real environment information on the positioning estimation capability based on the observation sensitivity and the preset basic uncertainty for the observation noise, wherein the observation sensitivity characterizes the sensitivity of the predicted environment information to the predicted positioning state; decompose the observation constraint matrix to determine the maximum constraint strength that the real environment information can provide for each state component in the positioning estimation; and determine the uncertainty of the observation noise based on the maximum constraint strength and the scene type in which the target object is located at the current moment, wherein the scene type includes isotropic scenes characterizing the consistent constraint strength of the real environment information in all directions, and anisotropic scenes characterizing the inconsistent constraint strength of the real environment information in all directions.

[0253] In some optional implementations of this embodiment, the observation weighting unit 502 is further configured to: in response to the target object being in an isotropic scene at the current moment, determine the adjustment weight for the basic uncertainty based on the maximum constraint strength; and combine the adjustment weight and the basic uncertainty to determine the uncertainty of the observation noise.

[0254] In some optional implementations of this embodiment, the observation weight unit 502 is further configured to: obtain an intensity metric by combining the maximum constraint strength corresponding to each state component; and determine the adjustment weight based on the normalized intensity metric and the preset metric threshold.

[0255] In some optional implementations of this embodiment, the observation weighting unit 502 is further configured to: in response to the target object being in an anisotropic scene at the current moment, determine a weight matrix representing the weight of each state component based on the maximum constraint strength corresponding to each state component; and determine the uncertainty of the observation noise based on the weight matrix, the observation sensitivity, and the observation constraint matrix, wherein the observation sensitivity represents the sensitivity of the change in the predicted environmental information at the current moment to the change in the predicted positioning state at the current moment.

[0256] In some optional implementations of this embodiment, the observation weight unit 502 is further configured to: for each state component, determine the weight corresponding to the state component based on the maximum constraint strength of the state component; and determine the weight matrix by combining the weights of each state component.

[0257] In some optional implementations of this embodiment, the observation weighting unit 502 is further configured to: adjust the maximum constraint strength corresponding to each state component according to the level of the maximum constraint strength corresponding to each state component, to obtain the adjusted observation constraint matrix of the built-in weight matrix; and determine the uncertainty of the observation noise according to the observation sensitivity and the adjusted observation constraint matrix.

[0258] In some optional implementations of this embodiment, the observation weight unit 502 is further configured to: divide each state component into at least one component group according to the comparison result of the preset strength threshold and the maximum constraint strength corresponding to each state component; adjust the maximum constraint strength corresponding to the state component in the component group by adopting the adjustment method corresponding to each component group, and determine the adjusted constraint strength of the weight corresponding to the built-in state component; and obtain the adjusted observation constraint matrix by combining the adjusted constraint strength corresponding to each state component.

[0259] In some optional implementations of this embodiment, the predicted positioning state is determined based on the final positioning state of the target object at the previous moment and the inertial data of the target object between the previous moment and the current moment. The device further includes: a data measurement unit (not shown in the figure), configured to determine the uncertainty of the predicted positioning state based on the uncertainty of the final positioning state of the target object at the previous moment, the state transition function, and the uncertainty of noise during the inertial data acquisition process, wherein the state transition function characterizes the correspondence between the final positioning state at the previous moment, the inertial data, and the predicted positioning state; and the positioning determination unit 503 is further configured to: adjust the predicted positioning state based on the difference, the uncertainty of the observation noise of the real environment information, the uncertainty of the predicted positioning state, and the observation sensitivity, and determine the final positioning state of the target object at the current moment, wherein the observation sensitivity characterizes the sensitivity of the predicted environment information to the predicted positioning state.

[0260] In some optional implementations of this embodiment, the data measurement unit is further configured to: determine state transition sensitivity and noise input sensitivity according to the state transition function, wherein the state transition sensitivity characterizes the sensitivity of the predicted positioning state to the final positioning state at the previous moment, and the noise input sensitivity characterizes the sensitivity of the predicted positioning state to noise during the inertial data acquisition process; and determine the uncertainty of the predicted positioning state based on the uncertainty of the final positioning state at the previous moment, the state transition sensitivity, the noise input sensitivity, and the uncertainty of noise during the inertial data acquisition process.

[0261] In some optional implementations of this embodiment, the positioning determination unit 503 is further configured to: determine the uncertainty of the difference based on the uncertainty of the observation noise, the uncertainty of the predicted positioning state, and the observation sensitivity; determine the correction weight of the real environment information relative to the predicted positioning state by combining the uncertainty of the difference, the observation sensitivity, and the uncertainty of the predicted positioning state; and determine the final positioning state at the current moment based on the predicted positioning state, the correction weight, and the difference.

[0262] In some optional implementations of this embodiment, the data measurement unit is further configured to determine the uncertainty of the final positioning state at the current moment based on the uncertainty of the predicted positioning state at the current moment, the correction weight, the observation sensitivity, and the uncertainty of the observation noise.

[0263] In some optional implementations of this embodiment, the above apparatus further includes: a validity determination unit (not shown in the figure), configured to determine the validity of the real environment information based on the difference and the uncertainty of the difference; and a positioning determination unit 503 further configured to: in response to the real environment information being a valid observation, adjust the predicted positioning state based on the difference, and determine the final positioning state of the target object at the current moment.

[0264] In this embodiment, a device for determining the positioning state is provided. An environmental prediction unit determines the predicted environmental information of the target object in the current time's predicted positioning state using an environmental prediction model. The environmental prediction model is used to characterize the correspondence between the positioning state and the environmental information. An observation weighting unit determines the uncertainty of the observation noise of the real environmental information based on the constraint strength of the target object's real environmental information on the positioning estimation capability at the current time. A positioning determination unit adjusts the predicted positioning state based on the uncertainty of the observation noise and the difference between the predicted environmental information and the real environmental information to determine the final positioning state of the target object at the current time. Thus, by combining the environmental prediction model and the inversion positioning method, the final positioning state is output as a state estimate, which has temporal continuity and improves the accuracy and precision of the positioning state. Furthermore, by dynamically determining the uncertainty of the observation noise of the real environmental information based on the constraint strength of the real environmental information on the positioning estimation capability, and adjusting the predicted positioning state accordingly to determine the final positioning state, the device can effectively suppress the negative impact of environmental signal degradation areas on the final positioning state, further improving the accuracy of the positioning state.

[0265] According to embodiments of this disclosure, this disclosure also provides an electronic device, the electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to implement the method for determining the positioning state described in any of the above embodiments when executed.

[0266] According to embodiments of this disclosure, this disclosure also provides a readable storage medium storing computer instructions that, when executed by a computer, enable the determination method for the positioning state described in any of the above embodiments.

[0267] This disclosure provides a computer program product that, when executed by a processor, can implement the method for determining the positioning state described in any of the above embodiments.

[0268] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0269] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded into random access memory (RAM) 603 from storage unit 608. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0270] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0271] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the method for determining the location state. For example, in some embodiments, the method for determining the location state may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the method for determining the location state described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the method for determining the location state by any other suitable means (e.g., by means of firmware).

[0272] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0273] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable positioning and state determination device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0274] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0275] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0276] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0277] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, also known as cloud computing servers or cloud hosts, which are hosting products within the cloud computing service system to address the management difficulties and weak business scalability inherent in traditional physical hosts and Virtual Private Servers (VPS) services; they can also be servers for distributed systems or servers incorporating blockchain technology.

[0278] According to the technical solution of the embodiments of this disclosure, a method and apparatus for determining the positioning state are provided. The method determines the predicted environmental information of a target object in its predicted positioning state at the current moment through an environmental prediction model, whereby the environmental prediction model characterizes the correspondence between the positioning state and the environmental information. Based on the constraint strength of the target object's actual environmental information on the positioning estimation capability at the current moment, the uncertainty of the observation noise of the actual environmental information is determined. Based on the uncertainty of the observation noise and the difference between the predicted environmental information and the actual environmental information, the predicted positioning state is adjusted to determine the final positioning state of the target object at the current moment. Thus, by combining the environmental prediction model and the inversion positioning method, the final positioning state is output as a state estimate, exhibiting temporal continuity and improving the accuracy and precision of the positioning state. Furthermore, by dynamically determining the uncertainty of the observation noise of the actual environmental information based on the constraint strength of the actual environmental information on the positioning estimation capability, and adjusting the predicted positioning state accordingly to determine the final positioning state, the method can effectively suppress the negative impact of environmental signal degradation areas on the final positioning state, further improving the accuracy of the positioning state.

[0279] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution provided in this disclosure can be achieved, and this is not limited herein.

[0280] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for determining a positioning state, comprising: The predicted environmental information of the target object in the predicted positioning state at the current moment is determined by an environmental prediction model, wherein the environmental prediction model is used to characterize the correspondence between the positioning state and the environmental information. The uncertainty of the observation noise of the real environment information is determined based on the constraint strength of the target object's real environment information on the positioning estimation capability at the current moment; Based on the uncertainty of the observed noise and the difference between the predicted environmental information and the actual environmental information, the predicted positioning state is adjusted to determine the final positioning state of the target object at the current moment.

2. The method according to claim 1, wherein, The step of determining the uncertainty of the observation noise of the real environment information based on the constraint strength of the positioning estimation capability on the real environment information of the target object at the current moment includes: Based on the observation sensitivity and the basic uncertainty preset for the observation noise, an observation constraint matrix is ​​determined to characterize the constraint strength of the real environment information on the positioning estimation capability, wherein the observation sensitivity characterizes the sensitivity of the predicted environment information to the predicted positioning state. Decompose the observation constraint matrix to determine the maximum constraint strength that the real environment information can provide for each state component in the positioning estimation; The uncertainty of the observation noise is determined based on the maximum constraint strength and the scene type in which the target object is located at the current moment. The scene type includes isotropic scenes that represent the consistent constraint strength of the real environment information in all directions, and anisotropic scenes that represent the inconsistent constraint strength of the real environment information in all directions.

3. The method according to claim 2, wherein, The step of determining the uncertainty of the observation noise based on the maximum constraint strength and the scene type in which the target object is located at the current moment includes: In response to the target object being in the isotropic scenario at the current moment, the adjustment weight for the basic uncertainty is determined based on the maximum constraint strength; The uncertainty of the observation noise is determined by combining the adjusted weights and the basic uncertainty.

4. The method according to claim 3, wherein, The step of determining the adjustment weight for the basic uncertainty based on the maximum constraint strength includes: By combining the maximum constraint strength corresponding to each state component, a strength metric is obtained; The adjustment weights are determined based on the normalized intensity metric and the preset metric threshold.

5. The method according to claim 2, wherein, The step of determining the uncertainty of the observation noise based on the maximum constraint strength and the scene type in which the target object is located at the current moment includes: In response to the target object being in the anisotropic scenario at the current moment, a weight matrix representing the weight of each state component is determined based on the maximum constraint strength corresponding to each state component. The uncertainty of the observation noise is determined based on the weight matrix, the observation sensitivity, and the observation constraint matrix, wherein the observation sensitivity characterizes the sensitivity of the change in the predicted environmental information at the current moment to the change in the predicted positioning state at the current moment.

6. The method according to claim 5, wherein, The step of determining the weight matrix representing the weight of each state component based on the maximum constraint strength corresponding to each state component includes: For each state component, the weight corresponding to the state component is determined based on the maximum constraint strength of the state component; The weight matrix is ​​determined by combining the weights of each state component.

7. The method according to claim 5, wherein, The step of determining the uncertainty of the observation noise based on the weight matrix, the observation sensitivity, and the observation constraint matrix includes: Based on the level of the maximum constraint strength corresponding to each state component, the maximum constraint strength corresponding to each state component is adjusted to obtain the adjusted observation constraint matrix with the built-in weight matrix. The uncertainty of the observation noise is determined based on the observation sensitivity and the adjusted observation constraint matrix.

8. The method according to claim 7, wherein, The step of adjusting the maximum constraint strength corresponding to each state component according to the level of the maximum constraint strength corresponding to each state component to obtain the adjusted observation constraint matrix with the built-in weight matrix includes: Based on the comparison results of the preset strength threshold and the maximum constraint strength corresponding to each state component, each state component is divided into at least one component group; By adopting the adjustment method corresponding to each component group, the maximum constraint strength corresponding to the state component in the component group is adjusted, and the adjusted constraint strength corresponding to the weight of the built-in state component is determined. The adjusted observation constraint matrix is ​​obtained by combining the adjusted constraint strengths corresponding to each state component.

9. The method according to any one of claims 1-8, wherein, The predicted positioning state is determined based on the target object's final positioning state at the previous moment, and the inertial data of the target object between the previous moment and the current moment. Also includes: The uncertainty of the predicted positioning state is determined based on the uncertainty of the target object's final positioning state at the previous moment, the uncertainty of the state transition function, and the uncertainty of noise during the inertial data acquisition process. The state transition function characterizes the correspondence between the final positioning state at the previous moment, the inertial data, and the predicted positioning state. The step of adjusting the predicted positioning state based on the uncertainty of the observed noise and the difference between the predicted environmental information and the actual environmental information, and determining the final positioning state of the target object at the current moment, includes: Based on the uncertainty of the observation noise, the difference, the uncertainty of the predicted positioning state, and the observation sensitivity, the predicted positioning state is adjusted to determine the final positioning state of the target object at the current moment, wherein the observation sensitivity characterizes the sensitivity of the predicted environmental information to the predicted positioning state.

10. The method according to claim 9, wherein, The step of determining the uncertainty of the predicted positioning state based on the uncertainty of the final positioning state of the target object at the previous moment, the uncertainty of the state transition function, and the uncertainty of noise during the inertial data acquisition process includes: Based on the state transition function, the state transition sensitivity and noise input sensitivity are determined, wherein the state transition sensitivity characterizes the sensitivity of the predicted positioning state to the final positioning state at the previous moment, and the noise input sensitivity characterizes the sensitivity of the predicted positioning state to noise during the inertial data acquisition process. The uncertainty of the predicted positioning state is determined based on the uncertainty of the final positioning state at the previous moment, the state transition sensitivity, the noise input sensitivity, and the uncertainty of noise during the inertial data acquisition process.

11. The method according to claim 9, wherein, The step of adjusting the predicted positioning state based on the uncertainty of the observation noise, the difference, the uncertainty of the predicted positioning state, and the observation sensitivity, to determine the final positioning state of the target object at the current moment, includes: The uncertainty of the difference is determined based on the uncertainty of the observation noise, the uncertainty of the predicted positioning state, and the observation sensitivity. By combining the uncertainty of the difference, the observation sensitivity, and the uncertainty of the predicted positioning state, a correction weight for the real environment information relative to the predicted positioning state is determined; The final positioning status at the current moment is determined based on the predicted positioning status, the correction weight, and the difference.

12. The method according to claim 9, wherein, Also includes: The uncertainty of the final positioning state at the current moment is determined based on the uncertainty of the predicted positioning state at the current moment, the correction weight, the observation sensitivity, and the uncertainty of the observation noise.

13. The method according to any one of claims 1-12, wherein, Also includes: The validity of the real-world environmental information is determined based on the difference and the uncertainty of the difference. as well as The step of adjusting the predicted positioning state based on the uncertainty of the observed noise and the difference between the predicted environmental information and the actual environmental information, and determining the final positioning state of the target object at the current moment, includes: In response to the fact that the real environment information is a valid observation, the predicted positioning state is adjusted according to the uncertainty of the observation noise and the difference, and the final positioning state of the target object at the current moment is determined.

14. A device for determining a positioning state, comprising: An environmental prediction unit is configured to determine the predicted environmental information of a target object in its current predicted positioning state through an environmental prediction model, wherein the environmental prediction model is used to characterize the correspondence between the positioning state and the environmental information. The observation weighting unit is configured to determine the uncertainty of the observation noise of the real environment information based on the constraint strength of the positioning estimation capability on the real environment information of the target object at the current moment. The positioning determination unit is configured to adjust the predicted positioning state based on the uncertainty of the observed noise and the difference between the predicted environmental information and the actual environmental information, and determine the final positioning state of the target object at the current moment.

15. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-13.

16. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-13.

17. A computer program product comprising: A computer program that, when executed by a processor, implements the method according to any one of claims 1-13.

Images