A positioning method and system based on multi-sensor fusion

Abstract

This application relates to the field of positioning technology, and in particular to a positioning method and system based on multi-sensor fusion. The method includes acquiring an initial environmental feature set and setting a dynamic sensing range; extracting feature points and collecting a synchronous sensing dataset; performing hierarchical analysis to obtain basic location information and environmental interference information to generate a comprehensive positioning model; using the model for preliminary calibration; obtaining a changing interference and response feature set through dynamic interference source excitation; plotting the original interference response curve and dividing the system into stable and fluctuating regions; completing dynamic interference compensation and plotting the compensated interference response curve; and finally achieving complete calibration of the positioning device. This application can optimize sensor data fusion strategies and improve the system's real-time performance, robustness, and adaptability to complex environments.

Description

Technical Field

[0001] This invention belongs to the field of positioning and navigation technology, specifically a positioning method and system based on multi-sensor fusion. Background Technology

[0002] With the rapid development of multi-sensor fusion positioning technology, positioning methods and systems based on the fusion of data from multiple sensors are increasingly widely used in navigation, robotics, and autonomous driving. These technologies, by integrating data from different sensors, can significantly improve positioning accuracy and robustness, meeting the high-precision positioning requirements in complex scenarios. However, existing multi-sensor fusion positioning methods still have certain shortcomings in terms of sensor data fusion accuracy, real-time performance, adaptability to complex environments, and system robustness, affecting the overall performance and applicability of the positioning system.

[0003] A search revealed a patent document, CN106595635B, which discloses an AGV positioning method that integrates data from multiple positioning sensors, published on December 8, 2020. This patent achieves high-precision AGV positioning by fusing data from three sensors: a laser sensor with reflective strips, a camera identification code strip, and feedback values ​​from a motor encoder. However, the priority setting of sensor data in this technical solution depends on specific scenarios, which may lead to a decrease in positioning accuracy in dynamic or complex environments. Furthermore, this method has high requirements for sensor calibration and initialization processes, increasing the complexity and implementation difficulty of the system and limiting its flexibility and universality in practical applications.

[0004] Another patent document, CN116295355B, discloses a mapping and localization method based on multi-sensor fusion, published on June 17, 2025. This patent achieves real-time mapping and localization of a mobile robot by combining three sensors: LocalSense, 2D LiDAR, and an RGB camera. However, this technical solution uses block processing and convolutional autoencoder compression when processing RGB images, which may introduce additional time delays, thus affecting the system's real-time performance. Furthermore, this method has high requirements for the synchronization of data from different sensors; inconsistencies in the timestamps of multi-sensor data may lead to increased localization errors, further weakening the system's robustness and reliability.

[0005] The aforementioned issues indicate that existing multi-sensor fusion positioning methods still have certain shortcomings in terms of adaptability, real-time performance, data synchronization, and system robustness in complex environments. Especially in dynamically changing environments, the uncertainty of sensor data, noise interference, and time synchronization problems can significantly affect positioning accuracy and system stability. Furthermore, the high dependence of existing technologies on sensor calibration and initialization increases system complexity and implementation difficulty, limiting their widespread application in real-world scenarios. Therefore, there is an urgent need for a novel multi-sensor fusion-based positioning method and system to optimize sensor data fusion strategies, improve system real-time performance and robustness, and enhance its adaptability in complex environments, thereby meeting the demands of modern positioning systems for high accuracy, high efficiency, and high reliability. Summary of the Invention

[0006] This invention provides a positioning method and system based on multi-sensor fusion, with the main objective of optimizing sensor data fusion strategies, improving system real-time performance and robustness, and enhancing its adaptability in complex environments. To achieve the above objectives, this invention provides a positioning method based on multi-sensor fusion, comprising: acquiring an initial environmental feature set of a target scene; setting a dynamic sensing range based on the initial environmental feature set; sequentially extracting feature points within the dynamic sensing range; collecting data from a pre-constructed multi-sensor integrated framework based on the feature points to obtain a synchronous sensing dataset; performing hierarchical analysis on the synchronous sensing dataset to obtain basic location information and environmental interference information; generating a comprehensive positioning model based on the basic location information and environmental interference information; using the comprehensive positioning model to perform preliminary position calibration on a pre-constructed positioning device to obtain a preliminary calibration device; and using a preset dynamic interference source... The multi-sensor integrated framework is continuously excited to obtain a set of changing interference features and a set of changing response features. Based on the set of changing interference features and the set of changing response features, an original interference response curve is plotted. The original interference response curve is divided into regions to obtain a set of stable regions and a set of fluctuating regions. The set of fluctuation factors of the fluctuating regions is identified. Based on the set of fluctuation factors and the set of stable regions, dynamic interference compensation is performed to obtain a set of compensated responses. The set of compensated interference corresponding to the set of compensated responses is identified. Based on the set of compensated responses and the set of compensated interference, a set of compensated interference response curves is plotted. The preliminary calibration device is then used to perform final position calibration using the set of compensated interference response curves to obtain a fully calibrated device.

[0007] Optionally, setting the dynamic sensing range based on the initial environmental feature set includes: setting the number of sensing divisions according to the initial environmental feature set; identifying the maximum and minimum environmental feature values ​​of the initial environmental feature set; calculating the division step size based on the number of sensing divisions, the maximum and minimum environmental feature values, to obtain the sensing division step size, wherein the sensing division step size is expressed as: dividing the initial environmental feature set into nodes based on the sensing division step size to obtain the dynamic sensing range.

[0008] Optionally, generating a comprehensive positioning model based on basic location information and environmental interference information includes: setting standard environmental conditions; determining the target position offset based on the standard environmental conditions, basic location information, and environmental interference information; measuring real-time environmental parameters of the multi-sensor integrated framework using a pre-built environmental monitoring module; setting correction weights for the real-time environmental parameters; identifying the core sensors within the multi-sensor integrated framework; determining sensor sensitivity and sensor error range based on the core sensors; querying environmental parameter benchmark values; and generating a comprehensive positioning model based on the environmental parameter benchmark values, sensor sensitivity, target position offset, standard environmental conditions, real-time environmental parameters, and correction weights.

[0009] Optionally, the step of using the integrated positioning model to perform preliminary position calibration on the pre-constructed positioning device to obtain a preliminary calibration device includes: calculating a set of position deviations between the integrated positioning model and the actual position information; sequentially extracting position deviations from the set of position deviations and determining whether the position deviations are greater than a preset standard deviation threshold; if the position deviations are greater than the standard deviation threshold, recording the position deviations as compensated position deviations, identifying the core sensor errors corresponding to the compensated position deviations, pairing the core sensor errors and the compensated position deviations to obtain a compensated position deviation group; summarizing the compensated position deviation groups to obtain a compensated position deviation set; and performing position deviation calibration on the positioning device based on the compensated position deviation set to obtain a preliminary calibration device.

[0010] Optionally, the step of continuously stimulating the multi-sensor integrated framework with a preset dynamic interference source to obtain a changing interference feature set and a changing response feature set includes: setting an initial excitation intensity and an initial excitation duration; stimulating the multi-sensor integrated framework with a dynamic interference source based on the initial excitation intensity and initial excitation duration to obtain an initial sensing dataset; extracting parameters from the initial sensing dataset according to a preset data sampling frequency and data sampling duration to obtain an initial interference feature set and an initial response feature set; calculating the average interference value and average response value of the initial interference feature set and the initial response feature set respectively; calculating the interference difference based on the average interference value and a preset historical interference value; determining whether the interference difference is greater than a preset interference threshold; if the interference difference is greater than the interference threshold, updating the historical interference value and the multi-sensor integrated framework with the average interference value and the initial sensing dataset respectively, and returning the step of continuously ... the preset historical interference value and the initial response feature set according to the initial interference intensity and initial excitation duration to obtain an initial sensing dataset; The steps are as follows: First, the dynamic interference source excites the multi-sensor integrated framework. If the interference difference is not greater than the interference threshold, the average interference value and average response value are recorded as the target interference value and target response value, respectively. The target interference value and target response value are used to supplement the pre-constructed initial variable interference set and initial variable response set to obtain the target variable interference set and target variable response set. The target sampling number of the target variable interference set is obtained, and it is determined whether the target sampling number is greater than the preset standard sampling number. If the target sampling number is not greater than the standard sampling number, the target variable interference set and target variable response set are used to update the initial variable interference set and initial variable response set, respectively, and the steps of exciting the multi-sensor integrated framework using a dynamic interference source based on the initial excitation intensity and initial excitation duration are returned. If the target sampling number is greater than the standard sampling number, the target variable interference set and target variable response set are recorded as the variable interference feature set and variable response feature set, respectively.

[0011] Optionally, the step of dividing the original interference response curve into regions to obtain a stable region set and a fluctuating region set includes: performing a preliminary artificial division of the original interference response curve to obtain a preliminary region set; performing the following operations on each preliminary region in the preliminary region set: identifying the boundary point set of the preliminary region, sequentially extracting boundary points from the boundary point set, identifying the adjacent boundary points of the boundary points, and calculating the boundary slope of the boundary points and their adjacent boundary points; summarizing the boundary slopes to obtain a boundary slope set, and calculating the regional fluctuation factor of the boundary slope set; determining whether the regional fluctuation factor is greater than a preset fluctuation threshold; if the regional fluctuation factor is not greater than the fluctuation threshold, then the preliminary region is recorded as a stable region; if the regional fluctuation factor is greater than the fluctuation threshold, then the preliminary region is recorded as a fluctuating region; summarizing the stable regions and the fluctuating regions respectively to obtain a stable region set and a fluctuating region set.

[0012] Optionally, the step of performing dynamic disturbance compensation based on the set of fluctuation factors and the set of stable regions to obtain a compensation response set includes: performing the following operations on each fluctuation factor in the set of fluctuation factors: identifying the fluctuation region to be processed corresponding to the fluctuation factor; extracting the center fluctuation point of the fluctuation region to be processed, and segmenting the fluctuation region to be processed based on the center fluctuation point to obtain a set of family fluctuation regions; sequentially extracting family fluctuation regions from the set of family fluctuation regions, identifying the family fluctuation point set of the family fluctuation regions, and calculating the family fluctuation factor of the family fluctuation point set; determining whether the family fluctuation factor is greater than the fluctuation threshold; if the family fluctuation factor is not greater than the fluctuation threshold, then recording the family fluctuation region as an equivalent stable region; if the family fluctuation factor is greater than the fluctuation threshold, then updating the fluctuation region to be processed using the family fluctuation region, and returning to the step of extracting the center fluctuation point of the fluctuation region to be processed; summarizing the equivalent stable regions to obtain an equivalent stable region set, supplementing the stable region set with the equivalent stable region set to obtain a set of stable regions to be compensated; and performing dynamic disturbance compensation based on the set of stable regions to be compensated to obtain a compensation response set.

[0013] Optionally, the step of performing dynamic interference compensation based on the set of stable regions to be compensated to obtain a compensation response set includes: performing the following operations on each stable region to be compensated in the set of stable regions to be compensated: extracting a central compensation point in the stable region to be compensated, and determining the central interference value and central response value of the central compensation point; determining the central compensation response value according to the standard environmental conditions, the central interference value, and the central response value; calculating a stability compensation coefficient based on the central compensation response value and the central response value; sequentially extracting points to be compensated in the stable region to be compensated, identifying the compensation response value of the points to be compensated, performing dynamic interference compensation according to the compensation response value and the stability compensation coefficient to obtain a compensation response value; and summarizing the compensation response values ​​to obtain a compensation response set.

[0014] Optionally, the step of using the compensation interference response curve to perform final position calibration on the preliminary calibration device to obtain a fully calibrated device includes: determining the control center of the preliminary calibration device, wherein the control center is an embedded control system; using the compensation interference response curve to input data to the control center to obtain a compensation control center; and connecting the compensation control center and the preliminary calibration device to obtain a fully calibrated device.

[0015] To achieve the above objectives, the present invention also provides a positioning system based on multi-sensor fusion, comprising: a perception data acquisition module, used to acquire an initial environmental feature set of a target scene, set a dynamic perception range based on the initial environmental feature set, sequentially extract feature points within the dynamic perception range, and acquire data from a pre-constructed multi-sensor integrated framework based on the feature points to obtain a synchronous perception dataset; a preliminary calibration module, used to perform hierarchical analysis on the synchronous perception dataset to obtain basic location information and environmental interference information, generate a comprehensive positioning model based on the basic location information and environmental interference information, and use the comprehensive positioning model to perform preliminary position calibration on a pre-constructed positioning device to obtain a preliminary calibration device; and an interference response rendering module, used to... The multi-sensor integrated framework is continuously excited by a preset dynamic interference source to obtain a changing interference feature set and a changing response feature set. Based on the changing interference feature set and the changing response feature set, an original interference response curve is plotted. The original interference response curve is divided into regions to obtain a stable region set and a fluctuating region set. The fluctuation factor set of the fluctuating region set is identified. Based on the fluctuation factor set and the stable region set, dynamic interference compensation is performed to obtain a compensation response set. The final calibration module is used to identify the compensation interference set corresponding to the compensation response set, plot the compensation interference response curve based on the compensation response set and the compensation interference set, and use the compensation interference response curve to perform final position calibration on the preliminary calibration device to obtain a fully calibrated device.

[0016] To address the aforementioned problems, the present invention also provides an electronic device, comprising: a memory storing at least one instruction; and a processor executing the instruction stored in the memory to implement the aforementioned multi-sensor fusion-based positioning method.

[0017] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in an electronic device to implement the aforementioned multi-sensor fusion-based positioning method.

[0018] To address the problems described in the background art, this invention obtains an initial environmental feature set of the target scene and sets a dynamic sensing range based on this set. Feature points are then extracted sequentially within the dynamic sensing range, defining the environmental feature range to be monitored and ensuring accurate collection and analysis of environmental data within this range. Next, data is collected from a multi-sensor integrated framework based on the feature points, resulting in a synchronous sensing dataset. This creates a synchronous sensing environment, facilitating accurate data analysis. The synchronous sensing dataset is then subjected to hierarchical analysis to obtain basic location information and environmental interference information. Based on this information, a comprehensive positioning model is generated, completing the construction of the comprehensive positioning model and providing a basis for subsequent position calibration. The comprehensive positioning model is used to perform preliminary position calibration on the positioning device, resulting in a preliminary calibration device. This step uses the comprehensive positioning model to calibrate the positioning device to compensate for potential positional deviations. By continuously exciting the multi-sensor integrated framework using dynamic interference sources, a changing interference feature set and a changing response feature set are obtained. Based on these features, the original interference response curve is plotted, providing an intuitive representation of the relationship between changing interference and changing response, thus improving calibration accuracy. Efficiency is improved, and due to the continuity of the curve, as long as the changing interference and response are on the original interference response curve and satisfy a linear relationship of a certain line segment, subsequent interference compensation can be performed even if the changing interference and response are not directly measured. This greatly enhances the adaptability of the positioning system. Next, the original interference response curve is divided into regions to obtain a stable region set and a fluctuating region set. By dividing the stable region set and the fluctuating region set, different changing interferences and responses can be calibrated separately, improving the accuracy and efficiency of calibration. Furthermore, the fluctuation factor set of the fluctuating region set is identified, and based on the fluctuation factor set... A stable region set is used to perform dynamic interference compensation, resulting in a compensation response set. This step completes the dynamic compensation for different interferences. By identifying the compensation interference set corresponding to the compensation response set, and based on the compensation response set and the compensation interference set, a compensation interference response curve is plotted. The plotting of the compensation interference response curve makes the required interference compensation more intuitive. At the same time, the curve form can also quickly locate the response value that needs to be compensated for different interferences, which improves the efficiency of the entire calibration. Finally, the compensation interference response curve is used to perform final position calibration on the preliminary calibration device, resulting in a fully calibrated device, thus completing the calibration of the positioning device. Therefore, this invention can optimize the sensor data fusion strategy, improve the system's real-time performance and robustness, and enhance its adaptability in complex environments. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a positioning method based on multi-sensor fusion provided in an embodiment of the present invention.

[0020] Figure 2 This is a functional block diagram of a positioning system based on multi-sensor fusion provided in an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the structure of an electronic device that implements the positioning method based on multi-sensor fusion, according to an embodiment of the present invention. Detailed Implementation

[0022] This invention provides a positioning method and system based on multi-sensor fusion. Its core lies in improving the system's real-time performance and robustness, and enhancing its adaptability in complex environments by optimizing the multi-sensor data fusion strategy. The following is in conjunction with the appendix... Figure 1 To be continued Figure 3 The specific embodiments of the present invention will be described in detail below.

[0023] First, in practical applications, obtaining the initial environmental feature set of the target scene is a fundamental step in the entire localization method. (See attached image) Figure 1 As shown, this process is completed through a sensing data acquisition module, which includes a set of high-precision sensors, such as LiDAR, cameras, and inertial measurement units (IMUs), used for preliminary scanning and feature extraction of the target scene. Assuming the target scene is an indoor warehouse, the initial environmental feature set may include the spatial distribution information of static objects such as walls, shelves, and the floor, as well as dynamic environmental parameters such as temperature and humidity. To ensure the accuracy of subsequent data acquisition, a dynamic sensing range needs to be set based on the initial environmental feature set. Specifically, by identifying the maximum and minimum environmental feature values ​​in the initial environmental feature set and combining them with a preset number of sensing partitions, the sensing partition step size is calculated. The formula is as follows: Sensing partition step size = (Maximum environmental feature value - Minimum environmental feature value) / Number of sensing partitions. Then, based on the sensing partition step size, the initial environmental feature set is divided into nodes to obtain the dynamic sensing range. For example, if the maximum environmental feature value is 50 meters, the minimum environmental feature value is 10 meters, and the number of sensing segments is 8, then the sensing segmentation step size is 5 meters, and the dynamic sensing range will be divided into 8 intervals: [10,15), [15,20), ..., [45,50]. These intervals define the areas that the sensor needs to focus on, thereby improving the targeting of data acquisition.

[0024] Next, feature points are extracted sequentially within the dynamic sensing range; this is a crucial step in generating the synchronous sensing dataset. (See attached image) Figure 1As shown, feature point extraction is performed by the perception data acquisition module. The specific process involves scanning sub-regions within each dynamic perception range using a multi-sensor integration framework. Taking LiDAR as an example, it determines the distance and angle of a target object by emitting a laser beam and receiving reflected signals, thereby generating a series of three-dimensional coordinate points. These points are called feature points, and they collectively constitute the synchronous perception dataset. The generation of the synchronous perception dataset relies on a multi-sensor time synchronization mechanism, meaning all sensors acquire data at the same timestamp, ensuring data consistency and reliability. For example, when LiDAR scans an edge of a shelf, a camera simultaneously captures an image of that edge, while the IMU records the device's attitude change information. These data are integrated to form the synchronous perception dataset, providing a foundation for subsequent analysis.

[0025] Subsequently, the synchronous sensing dataset was analyzed hierarchically to obtain basic location information and environmental interference information. (See attached image) Figure 1 As shown, this process is completed by the preliminary calibration module. The core idea of ​​hierarchical parsing is to decompose the synchronous sensing dataset into multiple layers, each corresponding to a different data type or feature. For example, the first layer may contain distance information, the second layer contains angle information, and the third layer contains interference information such as environmental noise. The extraction of basic position information mainly relies on geometric algorithms, such as triangulation or least squares methods. Taking triangulation as an example, assuming the positions of three fixed reference points are known as (x1, y1), (x2, y2), and (x3, y3), and the distances to these three reference points measured through the synchronous sensing dataset are d1, d2, and d3, then the target position (x, y) can be calculated using the following formula: (x - x1) 2 +(y-y1) 2 =d1 2 (x-x2) 2 +(y-y2) 2 =d2 2 (x-x3) 2 +(y-y3) 2 =d3 2 By solving this set of equations, the precise coordinates of the target location can be obtained. Meanwhile, the extraction of environmental interference information relies on statistical analysis methods, such as mean filtering or Kalman filtering. Taking Kalman filtering as an example, its state update equation is: X(k|k)=X(k|k-1)+K(k)·(Z(k)-H·X(k|k-1)), where X(k|k-1) is the predicted state, K(k) is the Kalman gain, Z(k) is the observed value, and H is the observation matrix. Through continuous iteration, noise can be effectively removed and the true environmental interference information can be extracted.

[0026] Generating a comprehensive positioning model based on basic location information and environmental interference information is one of the core components of the entire positioning method. (See attached image) Figure 1 As shown, this process is still completed by the preliminary calibration module. The generation of the integrated positioning model needs to consider multiple factors, including standard environmental conditions, real-time environmental parameters, sensor sensitivity, and error range. Specifically, standard environmental conditions are first set, such as a temperature of 25℃, humidity of 50%, and air pressure of 1013.25 hPa. Then, based on the standard environmental conditions, basic location information, and environmental interference information, the target position offset is determined. The formula for calculating the target position offset is: ΔP = P real -P standard , where P real P is the actual measured value of the target location. standard The standard value for the target location is given. Next, a pre-built environmental monitoring module is used to measure the real-time environmental parameters of the multi-sensor integrated framework, and correction weights for these parameters are set. For example, if the current temperature is 30℃ and the humidity is 60%, the correction weight might be 1.2. Furthermore, the core sensors within the multi-sensor integrated framework need to be identified, and their sensitivity and error range determined based on these core sensors. For example, the core sensor might be a lidar with a sensitivity of 0.01 meters and an error range of ±0.05 meters. Finally, the environmental parameter baseline value is queried, and a comprehensive positioning model is generated based on the environmental parameter baseline value, sensor sensitivity, target location offset, standard environmental conditions, real-time environmental parameters, and correction weights. The comprehensive positioning model can be expressed as: P final =P initial +ΔP*W correction , where P final For the final position, P initial As the initial position, W correction To adjust the weights.

[0027] The next crucial step is to perform preliminary position calibration of the positioning device using a comprehensive positioning model, which yields the initial calibration result. (See attached image.) Figure 1 As shown, this process is completed by the preliminary calibration module. The core of preliminary position calibration lies in calculating the set of position deviations between the integrated positioning model and the actual position information, and compensating for these deviations. Specifically, the set of position deviations between the integrated positioning model and the actual position information is first calculated. The formula for calculating the position deviation is: ΔD = P model -P actual , where P model For the location predicted by the integrated positioning model, P actualLet l represent the actual measured position. Then, position deviations are sequentially extracted from the position deviation set, and it is determined whether each deviation exceeds a preset standard deviation threshold. For example, if the standard deviation threshold is 0.1 meters, and a certain position deviation is 0.15 meters, then this position deviation is recorded as a compensated position deviation. Next, the core sensor error corresponding to the compensated position deviation is identified, and the core sensor error and the compensated position deviation are paired to obtain a compensated position deviation group. For example, if the compensated position deviation is 0.15 meters, and the core sensor error is 0.05 meters, then the compensated position deviation group is (0.15, 0.05). Finally, all compensated position deviation groups are summarized to obtain a compensated position deviation set, and the positioning device is calibrated based on the compensated position deviation set to obtain a preliminary calibration device.

[0028] Continuously exciting the multi-sensor integrated framework with preset dynamic interference sources to obtain changing interference feature sets and changing response feature sets is an important means to further improve positioning accuracy. (See attached image) Figure 1 As shown, this process is completed by the interference response rendering module. The dynamic interference source can be man-made vibration, electromagnetic interference, or other external disturbances. First, the initial excitation intensity and initial excitation duration are set. For example, the initial excitation intensity is 5N and the initial excitation duration is 10 seconds. Then, based on the initial excitation intensity and initial excitation duration, the multi-sensor integrated framework is excited using the dynamic interference source to obtain the initial sensing dataset. For example, when vibration interference is applied to the sensor, the sensor records the frequency, amplitude, and other relevant parameters of the vibration. Next, according to the preset data sampling frequency and data sampling duration, parameters are extracted from the initial sensing dataset to obtain the initial interference feature set and the initial response feature set. For example, if the data sampling frequency is 100Hz and the data sampling duration is 5 seconds, the initial interference feature set may contain interference data from 500 sampling points, and the initial response feature set contains the corresponding sensor response data. Subsequently, the average interference value and average response value of the initial interference feature set and the initial response feature set are calculated respectively, and the interference difference is calculated based on the average interference value and the preset historical interference value. The formula for calculating the interference difference is: ΔI = I average -I history , where I average I is the average interference value. historyThe historical interference values ​​are used. If the interference difference is greater than a preset interference threshold, the historical interference values ​​and the multi-sensor integration framework are updated using the average interference value and the initial sensing dataset, and the above excitation process is repeated. If the interference difference is not greater than the interference threshold, the average interference value and the average response value are recorded as the target interference value and the target response value, respectively. Finally, the pre-constructed initial variable interference set and initial variable response set are supplemented using the target interference value and the target variable response set, respectively, to obtain the target variable interference set and the target variable response set. If the target sampling number of the target variable interference set is greater than the preset standard sampling number, the target variable interference set and the target variable response set are recorded as the variable interference feature set and the variable response feature set, respectively.

[0029] Based on the changing interference characteristic set and the changing response characteristic set, the original interference response curve is plotted, and the original interference response curve is divided into regions to obtain the stable region set and the fluctuating region set. This is one of the key steps to achieve accurate calibration. (See attached image) Figure 1 As shown, this process is completed by the interference response plotting module. The original interference response curve is plotted using an interpolation algorithm, such as linear interpolation or cubic spline interpolation. Taking linear interpolation as an example, assume the changing interference feature set is {I1, I2, ..., I...} n The feature set of the change response is {R1, R2, ..., R}. n Then, any point (I,R) on the original interference response curve can be calculated using the following formula:

[0030]

[0031] Where I1≤I≤I2. After plotting, the original interference response curve is preliminarily divided to obtain a preliminary set of regions. For example, the curve is divided into 10 preliminary regions. Then, the following operations are performed on each preliminary region: identify the boundary point set of the preliminary region, extract the boundary points sequentially from the boundary point set, and identify the adjacent boundary points of the boundary points. For example, assuming the boundary points of a certain preliminary region are A and B, then the adjacent boundary points are A' and B'. Calculate the boundary slope of the boundary points and adjacent boundary points. The formula for calculating the boundary slope is:

[0032]

[0033] Where R A and R B Let I be the response values ​​at boundary points A and B, respectively. A and I B These are the disturbance values ​​for boundary points A and B, respectively. All boundary slopes are summarized to obtain a boundary slope set, and the regional volatility factor for this boundary slope set is calculated. The formula for calculating the regional volatility factor is: F = max(S set )-min(Sset ), where S set This represents the set of boundary slopes. If the regional volatility factor is not greater than a preset volatility threshold, the initially divided region is recorded as a stable region; otherwise, it is recorded as a volatile region. Finally, the stable regions and volatile regions are summarized separately to obtain the stable region set and the volatile region set.

[0034] Identifying the set of fluctuation factors within a set of fluctuating regions, and then performing dynamic disturbance compensation based on this set of fluctuation factors and the set of stable regions to obtain the compensation response set, is a crucial step in further optimizing the calibration results. (See attached...) Figure 1 As shown, this process is completed by the interference response rendering module. First, for each fluctuation factor in the fluctuation factor set, the following operations are performed: identify the fluctuation region to be processed corresponding to the fluctuation factor, and extract the center fluctuation point of the fluctuation region to be processed. For example, assuming the center fluctuation point of a certain fluctuation region is C, the fluctuation region to be processed is segmented with C as the center, resulting in a set of family-like fluctuation regions. Next, family-like fluctuation regions are extracted sequentially from the set of family-like fluctuation regions, and the set of family-like fluctuation points of the family-like fluctuation regions is identified. For example, assuming the set of family-like fluctuation points of a certain family-like fluctuation region is {C1, C2, ..., C...} m Then, calculate the family volatility factor of the family of volatility points. The formula for calculating the family volatility factor is:

[0035]

[0036] in This is the set of boundary slopes for the set of family-like fluctuation points. If the family-like fluctuation factor is not greater than the fluctuation threshold, the family-like fluctuation region is recorded as an equivalent stable region; otherwise, the family-like fluctuation region is used to update the fluctuation region to be processed, and the above segmentation process is repeated. Finally, all equivalent stable regions are summarized to obtain an equivalent stable region set, and the equivalent stable region set is used to supplement the stable region set to obtain a set of stable regions to be compensated. Based on the set of stable regions to be compensated, dynamic disturbance compensation is performed to obtain a compensation response set. Specifically, the following operations are performed for each stable region to be compensated: extract the center compensation point in the stable region to be compensated, and determine the center disturbance value and center response value of the center compensation point. For example, suppose the center disturbance value of a certain center compensation point is I. center The central response value is R center Based on the standard environmental conditions, the central interference value, and the central response value, the central compensation response value is determined. The formula for calculating the central compensation response value is: R comp =R center +(I center -I standard )·K adjust , where I standardLet K_adjust be the standard disturbance value and K_adjust be the adjustment coefficient. Next, based on the center compensation response value and the center response value, the stability compensation coefficient is calculated. The formula for calculating the stability compensation coefficient is: K_adjust = K_adjust + K_c. stable =R comp / R center Then, the points to be compensated are sequentially extracted within the stable region to be compensated, and the response values ​​to be compensated at each point are identified. For example, suppose the response value to be compensated at a certain point is R. target Then, based on the response value to be compensated and the stability compensation coefficient, dynamic disturbance compensation is performed to obtain the compensated response value. The formula for calculating the compensated response value is: R final =R target ·K stable Finally, all compensation response values ​​are summed to obtain the compensation response set.

[0037] Identifying the compensation interference set corresponding to the compensation response set, and plotting the compensation interference response curve based on the compensation response set and the compensation interference set, is a crucial step in achieving final calibration. (See attached image) Figure 1 As shown, this process is completed by the final calibration module. The plotting of the compensation interference response curve also employs an interpolation algorithm, such as linear interpolation or cubic spline interpolation. Taking linear interpolation as an example, assume the compensation response set is {R}. c1 ,R c2 ,...,R cn}, the compensation interference set is {I} c1 ,I c2 ,...,I cn Then, any point (I,R) on the compensation interference response curve can be calculated using the following formula:

[0038]

[0039] Among them I c1 ≤I≤I c2 After the plotting is complete, the preliminary calibration equipment is finally calibrated using the compensated interference response curve to obtain a fully calibrated equipment. Specifically, the control center of the preliminary calibration equipment is first determined, where the control center is an embedded control system. Then, the compensated interference response curve is used to input data into the control center to obtain the compensated control center. For example, suppose a point on the compensated interference response curve is (I... comp ,R comp ), then R comp The compensation value is input to the control center. Finally, the compensation control center and the preliminary calibration equipment are connected to obtain the fully calibrated equipment.

[0040] As attached Figure 2As shown, this invention also provides a positioning system based on multi-sensor fusion, including a sensing data acquisition module, a preliminary calibration module, an interference response drawing module, and a final calibration module. The sensing data acquisition module is responsible for acquiring the initial environmental feature set of the target scene, setting a dynamic sensing range based on the initial environmental feature set, sequentially extracting feature points within the dynamic sensing range, and collecting data from the multi-sensor integrated framework based on the feature points to obtain a synchronous sensing dataset. The preliminary calibration module is responsible for performing layered analysis on the synchronous sensing dataset to obtain basic location information and environmental interference information, generating a comprehensive positioning model based on the basic location information and environmental interference information, and using the comprehensive positioning model to perform preliminary position calibration on the positioning device to obtain a preliminary calibration device. The interference response drawing module is responsible for continuously stimulating the multi-sensor integrated framework through a preset dynamic interference source to obtain a changing interference feature set and a changing response feature set, drawing the original interference response curve based on the changing interference feature set and the changing response feature set, dividing the original interference response curve into regions to obtain a stable region set and a fluctuating region set, identifying the fluctuation factor set of the fluctuating region set, and performing dynamic interference compensation based on the fluctuation factor set and the stable region set to obtain a compensated response set. The final calibration module is responsible for identifying the compensation interference set corresponding to the compensation response set, and plotting the compensation interference response curve based on the compensation response set and the compensation interference set. The compensation interference response curve is then used to perform the final position calibration of the preliminary calibration equipment to obtain the fully calibrated equipment.

[0041] As attached Figure 3 As shown, the present invention also provides an electronic device, including a memory and a processor. The memory is used to store at least one instruction, and the processor is used to execute the instruction stored in the memory to implement the aforementioned positioning method based on multi-sensor fusion. Furthermore, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in the electronic device to implement the aforementioned positioning method based on multi-sensor fusion.

[0042] In summary, this invention acquires an initial environmental feature set of the target scene and sets a dynamic sensing range based on this set. Feature points are then extracted sequentially within the dynamic sensing range, defining the environmental feature range to be monitored and ensuring accurate collection and analysis of environmental data within this range. Next, data is collected from the multi-sensor integrated framework based on the feature points, resulting in a synchronous sensing dataset. This creates a synchronous sensing environment, facilitating accurate data analysis. Then, the synchronous sensing dataset is analyzed hierarchically to obtain basic location information and environmental interference information. Based on this information, a comprehensive positioning model is generated, completing the construction of the comprehensive positioning model and providing a basis for subsequent position calibration. The comprehensive positioning model is used to perform preliminary position calibration on the positioning device, resulting in a preliminary calibration device. This step uses the comprehensive positioning model to calibrate the positioning device to compensate for potential positional deviations. By continuously exciting the multi-sensor integrated framework with dynamic interference sources, a changing interference feature set and a changing response feature set are obtained. Based on these sets, the original interference response curve is plotted, providing an intuitive representation of the relationship between changing interference and changing response, thus improving calibration efficiency. Meanwhile, due to the continuity of the curves, as long as the changing interference and response lie on the original interference response curve and satisfy a linear relationship of a certain line segment, subsequent interference compensation can be performed even if the changing interference and response are not directly measured. This greatly enhances the adaptability of the positioning system. Next, the original interference response curve is divided into regions, resulting in a stable region set and a fluctuating region set. By dividing the stable and fluctuating regions, different changing interferences and responses can be calibrated separately, improving the accuracy and efficiency of the calibration. Further, the fluctuation factor set of the fluctuating region set is identified, and dynamic interference compensation is performed based on the fluctuation factor set and the stable region set to obtain the compensation response set. This step completes the dynamic compensation for different interferences. By identifying the compensation interference set corresponding to the compensation response set, and plotting the compensation interference response curve based on the compensation response set and the compensation interference set, the required interference compensation becomes more intuitive. Simultaneously, the curve format allows for quick location of the response values ​​that need compensation for different interferences, improving the overall calibration efficiency. Finally, the compensation interference response curve is used to perform final position calibration on the preliminary calibration equipment, resulting in a fully calibrated equipment, thus completing the calibration of the positioning equipment. Therefore, the present invention can optimize sensor data fusion strategies, improve system real-time performance and robustness, and enhance its adaptability in complex environments.

Claims

1. A positioning method based on multi-sensor fusion, characterized in that, The method includes: Obtain an initial environmental feature set of the target scene, set a dynamic perception range based on the initial environmental feature set, and extract feature points sequentially within the dynamic perception range; Based on the aforementioned feature points, data is collected from the pre-constructed multi-sensor integration framework to obtain a synchronous sensing dataset; The synchronous sensing dataset is analyzed hierarchically to obtain basic location information and environmental interference information. Based on the basic location information and environmental interference information, a comprehensive positioning model is generated. Using the integrated positioning model, the pre-constructed positioning device is initially calibrated to obtain a pre-calibrated device; By continuously exciting the multi-sensor integrated framework with a preset dynamic interference source, a set of changing interference features and a set of changing response features are obtained. Based on the changing interference feature set and the changing response feature set, the original interference response curve is plotted, and the original interference response curve is divided into regions to obtain the stable region set and the fluctuating region set. Identify the set of fluctuation factors of the set of fluctuation regions, and perform dynamic disturbance compensation based on the set of fluctuation factors and the set of stable regions to obtain a compensation response set; Identify the compensation interference set corresponding to the compensation response set, and plot the compensation interference response curve based on the compensation response set and the compensation interference set; The preliminary calibration equipment is then subjected to final position calibration using the compensation interference response curve to obtain a fully calibrated equipment.

2. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The step of setting the dynamic perception range based on the initial environmental feature set includes: Based on the initial environmental feature set, the number of perception segments is set; Identify the maximum and minimum environmental feature values ​​of the initial environmental feature set, and calculate the segmentation step size based on the number of perception segments, the maximum and minimum environmental feature values ​​to obtain the perception segmentation step size; Based on the perception partitioning step size, the initial environmental feature set is partitioned into nodes to obtain the dynamic perception range.

3. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The process of generating a comprehensive positioning model based on basic location information and environmental interference information includes: Set standard environmental conditions, and determine the target position offset based on the standard environmental conditions, basic location information, and environmental interference information; The real-time environmental parameters of the multi-sensor integrated framework are measured using a pre-built environmental monitoring module. Correction weights for the real-time environmental parameters are set, the core sensors within the multi-sensor integrated framework are identified, and the sensor sensitivity and sensor error range are determined based on the core sensors. The system queries environmental parameter baseline values ​​and generates a comprehensive positioning model based on these baseline values, sensor sensitivity, target position offset, standard environmental conditions, real-time environmental parameters, and correction weights.

4. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The step of using the integrated positioning model to perform preliminary position calibration on the pre-constructed positioning device to obtain a preliminary calibration device includes: Calculate the set of positional deviations between the integrated positioning model and the actual position information; The position deviations are extracted sequentially from the set of position deviations, and it is determined whether the position deviations are greater than a preset standard deviation threshold. If the position deviation is greater than the standard deviation threshold, the position deviation is recorded as the compensated position deviation, the core sensor error corresponding to the compensated position deviation is identified, and the core sensor error and the compensated position deviation are paired to obtain a compensated position deviation group. The compensation position deviation groups are summarized to obtain the compensation position deviation set. Based on the compensation position deviation set, the positioning device is calibrated to obtain the preliminary calibration device.

5. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The step of continuously exciting the multi-sensor integrated framework through a preset dynamic interference source to obtain a changing interference feature set and a changing response feature set includes: Set the initial stimulus intensity and initial stimulus duration; Based on the initial excitation intensity and initial excitation duration, the multi-sensor integrated framework is excited by a dynamic interference source to obtain the initial sensing dataset. Based on the preset data sampling frequency and data sampling duration, parameters are extracted from the initial sensing dataset to obtain the initial interference feature set and the initial response feature set; Calculate the average interference value and average response value of the initial interference feature set and the initial response feature set respectively. Calculate the interference difference based on the average interference value and the preset historical interference value, and determine whether the interference difference is greater than the preset interference threshold. If the interference difference is greater than the interference threshold, the historical interference value and the multi-sensor integrated framework are updated using the average interference value and the initial sensing dataset, and the step of stimulating the multi-sensor integrated framework using a dynamic interference source based on the initial excitation intensity and initial excitation duration is returned. If the interference difference is not greater than the interference threshold, the average interference value and the average response value are recorded as the target interference value and the target response value, respectively. The target interference value and target response value are used to supplement the pre-constructed initial change interference set and initial change response set to obtain the target change interference set and target change response set; Obtain the target sampling number of the target variation interference set, and determine whether the target sampling number is greater than the preset standard sampling number; If the number of target samples is not greater than the number of standard samples, the initial change interference set and the initial change response set are updated using the target change interference set and the target change response set respectively, and the step of stimulating the multi-sensor integrated framework with dynamic interference source based on the initial excitation intensity and the initial excitation duration is returned. If the number of target samples is greater than the number of standard samples, then the target change interference set and the target change response set are respectively denoted as the change interference feature set and the change response feature set.

6. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The process of dividing the original interference response curve into regions to obtain a stable region set and a fluctuating region set includes: The original interference response curve is preliminarily divided by human intervention to obtain a preliminary set of regions; For each initially partitioned region in the aforementioned initially partitioned region set, perform the following operations: Identify the set of boundary points of the initially divided region, extract the boundary points sequentially from the set of boundary points, identify the adjacent boundary points of the boundary points, and calculate the boundary slope of the boundary points and their adjacent boundary points; Summarize the boundary slopes to obtain the boundary slope set, and calculate the regional fluctuation factor of the boundary slope set. Determine whether the regional volatility factor is greater than a preset volatility threshold; If the regional volatility factor is not greater than the volatility threshold, then the initially divided region is recorded as a stable region; If the regional volatility factor is greater than the volatility threshold, then the initially divided region is recorded as a volatility region; By summarizing the stable regions and the fluctuating regions respectively, we obtain the stable region set and the fluctuating region set.

7. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The step of performing dynamic disturbance compensation based on the set of fluctuation factors and the set of stable regions to obtain a compensation response set includes: Perform the following operations on each volatility factor in the set of volatility factors: Identify the fluctuation region to be processed corresponding to the fluctuation factor; Extract the center fluctuation point of the fluctuation region to be processed, and segment the fluctuation region to be processed based on the center fluctuation point to obtain a set of fluctuation regions of the same family; Extract the same family fluctuation regions sequentially from the same family fluctuation region set, identify the same family fluctuation point set of the same family fluctuation region, and calculate the same family fluctuation factor of the same family fluctuation point set; Determine whether the homogeneous volatility factor is greater than the volatility threshold; If the family-like volatility factor is not greater than the volatility threshold, then the family-like volatility region is recorded as the equivalent stable region; If the family-based fluctuation factor is greater than the fluctuation threshold, then the family-based fluctuation region is used to update the fluctuation region to be processed, and the step of extracting the center fluctuation point of the fluctuation region to be processed is returned. The equivalent stable regions are summarized to obtain an equivalent stable region set. The stable region set is supplemented by the equivalent stable region set to obtain a stable region set to be compensated. Based on the set of stable regions to be compensated, dynamic disturbance compensation is performed to obtain a compensation response set.

8. The positioning method based on multi-sensor fusion as described in claim 7, characterized in that, The step of performing dynamic disturbance compensation based on the set of stable regions to be compensated to obtain a compensation response set includes: For each stable region to be compensated in the set of stable regions to be compensated, perform the following operations: Extract the central compensation point from the stable region to be compensated, and determine the central interference value and central response value of the central compensation point; The central compensation response value is determined based on the standard environmental conditions, the central interference value, and the central response value. Based on the central compensation response value and the central response value, calculate the stability compensation coefficient; In the stable region to be compensated, points to be compensated are extracted sequentially, the response values ​​to be compensated for the points to be compensated are identified, and dynamic disturbance compensation is performed based on the response values ​​to be compensated and the stability compensation coefficient to obtain the compensation response value. The compensation response values ​​are summed to obtain the compensation response set.

9. The positioning method based on multi-sensor fusion as described in claim 1, characterized in that, The step of performing final position calibration on the preliminary calibration equipment using the compensation interference response curve to obtain a fully calibrated equipment includes: The control center of the preliminary calibration equipment is determined, wherein the control center is an embedded control system; Using the compensation interference response curve, data is input to the control center to obtain the compensation control center. Connecting the compensation control center and the preliminary calibration equipment yields the fully calibrated equipment.

10. A positioning system based on multi-sensor fusion, characterized in that, The system includes: The perception data acquisition module is used to acquire the initial environmental feature set of the target scene, set the dynamic perception range based on the initial environmental feature set, extract feature points sequentially within the dynamic perception range, and collect data from the pre-constructed multi-sensor integrated framework based on the feature points to obtain a synchronous perception dataset. The preliminary calibration module is used to perform hierarchical analysis on the synchronous sensing dataset to obtain basic location information and environmental interference information. Based on the basic location information and environmental interference information, a comprehensive positioning model is generated. Using the comprehensive positioning model, the pre-constructed positioning device is initially calibrated to obtain a preliminary calibration device. The interference response plotting module is used to continuously excite the multi-sensor integrated framework through a preset dynamic interference source to obtain a changing interference feature set and a changing response feature set. Based on the changing interference feature set and the changing response feature set, the module plots the original interference response curve, divides the original interference response curve into regions to obtain a stable region set and a fluctuating region set, identifies the fluctuation factor set of the fluctuating region set, and performs dynamic interference compensation based on the fluctuation factor set and the stable region set to obtain a compensation response set. The final calibration module is used to identify the compensation interference set corresponding to the compensation response set, draw the compensation interference response curve based on the compensation response set and the compensation interference set, and use the compensation interference response curve to perform final position calibration on the preliminary calibration device to obtain a fully calibrated device.

Images