A spatio-temporal synchronization multi-sensor error analysis method and apparatus

Abstract

The application discloses a kind of space-time synchronization multi-sensor error analysis method and device, method includes: obtaining roadside pre-deployed thing networking sensor to the original data collected by passing vehicle and pre-processing;The data after pre-processing is converted from thing networking sensor coordinate system to global coordinate system;Establish space-time synchronization error model;Wherein, space-time synchronization error model is used to characterize the difference of position obtained by geomagnetic sensor and millimeter wave radar to the same vehicle at the same time;Space-time synchronization error model is composed of time synchronization error, space synchronization error and sensor measurement error three parts;Based on the data after coordinate conversion obtained by all thing networking sensors, sensor measurement error is solved respectively according to different vehicle types;Based on the sensor measurement error solved by different vehicle types, the space synchronization error of different vehicle types is obtained by fitting solution method.The application can solve the correlation error problem caused by space-time synchronization error of the data of different sensor devices.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology, specifically relating to a spatiotemporal synchronous multi-sensor error analysis method and device. Background Technology

[0002] Transportation is the driving force of urban development, but the ever-increasing number of cars has led to severe traffic congestion, seriously affecting urban socio-economic development. Intelligent Traffic Systems (ITS) effectively integrate computer technology, sensor technology, information technology, and communication technology into traffic information management, constructing a large-scale, comprehensive, accurate, real-time, and efficient integrated transportation management system, representing the future direction of transportation system development for decades to come. Diverse Internet of Things (IoT) sensors, such as cameras, geomagnetic sensors, and millimeter-wave radar, are deployed in transportation systems, providing accurate road and vehicle information and promoting the development of ITS. However, data fusion between different sensor devices, if directly using their temporal and spatial information for correlation and fusion, is severely affected by spatiotemporal synchronization errors. When these errors are large, correlation errors are prone to occur. Therefore, accurately determining and optimizing spatiotemporal synchronization errors is particularly important.

[0003] Tongji University proposed an automatic online calibration method for intelligent vehicles using multi-sensor spatiotemporal fusion in its patent application, "An Automatic Online Calibration Method for Multi-Sensor Spatiotemporal Fusion of Intelligent Vehicles" (Application No.: CN202210785604.6, Publication No.: CN115144828A). This method achieves automatic online calibration of intelligent vehicles by using spatiotemporal fusion of data from multiple sensors. First, the method collects vehicle operation data through various sensors, including information such as vehicle position, speed, and direction. Then, the collected data undergoes preprocessing, including data cleaning and data fusion, to improve data quality and accuracy. Next, a spatiotemporal fusion model is constructed to fuse the preprocessed data to obtain more comprehensive and accurate vehicle status information. Finally, online calibration is performed to adjust and optimize model parameters in real time to further improve the accuracy and stability of the calibration. The drawback of this method is that it involves the fusion of multiple sensors and real-time online calibration, which compromises the stability and reliability of the system. Summary of the Invention

[0004] To address the aforementioned problems in the existing technology, this invention provides a spatiotemporal synchronized multi-sensor error analysis method and apparatus. The technical problem to be solved by this invention is achieved through the following technical solution:

[0005] In a first aspect, embodiments of the present invention provide a spatiotemporal synchronized multi-sensor error analysis method, the method comprising:

[0006] The system acquires raw data collected by roadside IoT sensors from passing vehicles and performs preprocessing; the IoT sensors include geomagnetic sensors and millimeter-wave radar.

[0007] The preprocessed data is transformed from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data.

[0008] A spatiotemporal synchronization error model is established; wherein, the spatiotemporal synchronization error model is used to characterize the difference in position acquired by the geomagnetic sensor and the millimeter-wave radar of the same vehicle at the same moment; the spatiotemporal synchronization error model consists of three parts: time synchronization error, spatial synchronization error, and sensor measurement error;

[0009] Based on the coordinate transformation data obtained from all IoT sensors, the measurement errors of the sensors are solved separately for different vehicle models;

[0010] Based on the sensor measurement errors obtained from different vehicle models, the spatial synchronization error of different vehicle models is obtained by fitting the solution.

[0011] In one embodiment of the present invention, the process of pre-deploying IoT sensors on the roadside includes:

[0012] Multiple geomagnetic sensors are buried at a first preset interval on the road to the right of the vehicle's direction of travel;

[0013] Multiple millimeter-wave radars are installed at a second preset interval on the road to the right of the vehicle's direction of travel; wherein the receiving surface of the millimeter-wave radar faces the vehicle's direction of travel; and the height of the millimeter-wave radar above the ground is within a preset height range.

[0014] In one embodiment of the present invention, the preset height range is 4m to 6m.

[0015] In one embodiment of the present invention, the step of acquiring and preprocessing the raw data collected by roadside IoT sensors from passing vehicles includes:

[0016] The disturbance signal carrying a timestamp is collected when the geomagnetic sensor detects the passing of a vehicle. After filtering the disturbance signal, the timestamp of the peak of the disturbance signal is selected as the time when the vehicle passes the geomagnetic sensor.

[0017] The system acquires point cloud data with timestamps obtained by converting the reflected signals from the vehicle received by the millimeter-wave radar. It then clusters the point cloud data to generate vehicle trajectory data, thereby obtaining the vehicle's position, speed, and time.

[0018] In one embodiment of the present invention, the step of transforming the preprocessed data from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data includes:

[0019] Based on the actual road conditions and the installation locations of each IoT sensor, a reference coordinate system is selected as a unified global coordinate system, and the position of each IoT sensor in the global coordinate system is determined.

[0020] Based on the position of each IoT sensor in the global coordinate system, establish the corresponding coordinate transformation relationship from the IoT sensor coordinate system to the global coordinate system;

[0021] For each IoT sensor, based on the coordinate transformation relationship from its own IoT sensor coordinate system to the global coordinate system, its preprocessed data is transformed to the global coordinate system to obtain the coordinate-transformed data corresponding to that IoT sensor.

[0022] In one embodiment of the present invention, establishing the spatiotemporal synchronization error model includes:

[0023] Based on the time synchronization error caused by the asynchrony between the geomagnetic sensor time axis and the millimeter-wave radar time axis, and the vehicle speed, the expression for the time synchronization error is determined as follows:

[0024] δ(t) = v*dt;

[0025] Where v is the vehicle speed; d(t) is the time synchronization error, d(t) = t b -t r ;t b and t r These are the timestamps from the geomagnetic sensor and the millimeter-wave radar in the coordinate-transformed data, respectively, for the same location and the same vehicle.

[0026] Based on the difference between the installation error of the geomagnetic sensor and the installation error of the millimeter-wave radar, the expression for the space synchronization error is determined as follows:

[0027] δ(s)=ds b -ds r ;

[0028] ds b =s b -s' b ;

[0029] ds r =s r -s' r ;

[0030] Among them, ds bs represents the difference between the theoretical and actual installation locations of the geomagnetic sensor. b This represents the theoretical installation location of the geomagnetic sensor; s' b This refers to the actual installation location of the geomagnetic sensor; ds r s represents the difference between the theoretical and actual installation locations of the millimeter-wave radar; r This refers to the theoretical installation location of the millimeter-wave radar; s' r This refers to the actual installation location of the millimeter-wave radar.

[0031] Based on the fact that the actual measurement point measured by the millimeter-wave radar is the closest point on the vehicle to the millimeter-wave radar rather than the center point of the vehicle, and based on the possibility that the center point of the vehicle and the geomagnetic sensor are not aligned during the peak time of the disturbance signal measured by the geomagnetic sensor, the distribution characteristics of the sensor measurement error are determined to be: following a Gaussian distribution with a mean of half the length of the vehicle body.

[0032] Based on the time synchronization error, the spatial synchronization error, and the sensor measurement error, the expression for the spatiotemporal synchronization error model is established as follows:

[0033] d s =δ(t) + δ(s) + δ(m);

[0034] Where δ(t) is the time synchronization error; δ(s) is the spatial synchronization error; and δ(m) is the sensor measurement error.

[0035] In one embodiment of the present invention, the step of solving for the sensor measurement error based on the coordinate transformation data obtained from all IoT sensors, according to different vehicle models, includes:

[0036] Based on the distribution characteristics of the sensor measurement errors, the coordinate transformation data obtained from all IoT sensors are categorized into three types according to vehicle length: large vehicles, medium vehicles, and small vehicles. The general expression for the vehicle type of the sensor measurement errors is determined as follows:

[0037] δ(m)=A*δ(m) big +B*δ(m) middle +C*δ(m) small ;

[0038] Where, δ(m) big For large vehicles, the sensor measurement error is δ(m). middle The sensor measurement error for a medium-sized vehicle; δ(m) small The sensor measurement error is for a small car; A, B, and C represent δ(m) respectively. big δ(m) middle δ(m) smallThe corresponding proportional coefficients; if the vehicle is a large vehicle, A=1, B=0, C=0 in the general expression for vehicle type; if the vehicle is a medium-sized vehicle, A=0, B=1, C=0 in the general expression for vehicle type; if the vehicle is a small vehicle, A=0, B=0, C=1 in the general expression for vehicle type.

[0039] Based on the expression of the spatiotemporal synchronization error model, the calculation expressions for the spatiotemporal synchronization error models corresponding to large vehicles, medium-sized vehicles, and small vehicles are constructed respectively as follows:

[0040] d' s =v'*dt+δ(s)+δ(m) big ;

[0041] d” s =v”*dt+δ(s)+δ(m) middle ;

[0042] d”' s =v”'*dt+δ(s)+δ(m) small ;

[0043] Where v', v” and v”' represent the speeds of large vehicles, medium vehicles and small vehicles, respectively;

[0044] The difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and medium-sized vehicles is obtained to get the first expression; the difference between the spatiotemporal synchronization error model calculation expressions corresponding to medium-sized vehicles and small vehicles is obtained to get the second expression; the difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and small vehicles is obtained to get the third expression; wherein, the first expression, the second expression, and the third expression are respectively:

[0045] d' s -d” s = (v'-v”)*dt+(δ(m) big -δ(m) middle );

[0046] d” s -d”' s = (v”-v”')*dt+(δ(m) middle -δ(m) small );

[0047] d' s -d”' s = (v'-v”')*dt+(δ(m) big -δ(m) small );

[0048] The spatiotemporal synchronization error data and speed of all large, medium and small vehicles are obtained from the coordinate transformation data of all IoT sensors and merged into a dataset.

[0049] Based on the data of each vehicle type in the dataset, the first expression, the second expression, and the third expression are fitted and solved to obtain the sensor measurement error and time synchronization error corresponding to large vehicles, medium vehicles, and small vehicles, respectively.

[0050] In one embodiment of the present invention, the step of obtaining the spatial synchronization error of different vehicle models by fitting the sensor measurement errors obtained based on different vehicle models includes:

[0051] For each vehicle model, based on the sensor measurement error and time synchronization error obtained under that model, as well as the spatiotemporal synchronization error data and speed of each vehicle under that model, the spatial synchronization error of that model is obtained by fitting and solving the spatiotemporal synchronization error model calculation expression.

[0052] In one embodiment of the present invention, the fitting solution is achieved using the least squares method.

[0053] Secondly, embodiments of the present invention provide a spatiotemporal synchronized multi-sensor error analysis device, the device comprising:

[0054] The vehicle data acquisition and preprocessing module is used to acquire and preprocess raw data collected by roadside IoT sensors from passing vehicles; wherein, the IoT sensors include geomagnetic sensors and millimeter-wave radar.

[0055] The coordinate transformation module is used to transform the preprocessed data from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data;

[0056] A spatiotemporal synchronization error model establishment module is used to establish a spatiotemporal synchronization error model; wherein, the spatiotemporal synchronization error model is used to characterize the difference in position acquired by the geomagnetic sensor and the millimeter-wave radar of the same vehicle at the same moment; the spatiotemporal synchronization error model consists of three parts: time synchronization error, spatial synchronization error, and sensor measurement error;

[0057] The sensor measurement error solving module is used to solve the sensor measurement error separately according to different vehicle models based on the coordinate transformation data obtained from all IoT sensors.

[0058] The spatial synchronization error solving module is used to obtain the spatial synchronization error of different vehicle models by fitting the sensor measurement errors obtained from different vehicle models.

[0059] The beneficial effects of this invention are:

[0060] This invention provides a spatiotemporal synchronization multi-sensor error analysis method to address the error problem existing in multi-sensor data fusion in intelligent transportation systems. By establishing a spatiotemporal synchronization model among multiple sensors, this invention performs spatiotemporal synchronization processing on the data from multiple sensors to obtain sensor measurement errors and spatial synchronization errors corresponding to each vehicle model. These errors are then used for subsequent parameter compensation, correcting the system's spatiotemporal synchronization error and resolving the temporal and spatial alignment problem between different sensors. This improves the accuracy and reliability of the data, achieves precise calibration and optimization of sensor data, and solves the problem of association errors caused by spatiotemporal synchronization errors in data from different sensor devices, thus optimizing subsequent data association. Attached Figure Description

[0061] Figure 1 This is a flowchart illustrating a spatiotemporal synchronous multi-sensor error analysis method provided in an embodiment of the present invention;

[0062] Figure 2 This is a schematic diagram of the global coordinate system and the installation positions of each IoT sensor in an embodiment of the present invention;

[0063] Figure 3 This is a schematic diagram of spatial synchronization error in an embodiment of the present invention;

[0064] Figure 4 This is a schematic diagram of sensor measurement error in an embodiment of the present invention;

[0065] Figure 5 This is a schematic diagram of a spatiotemporal synchronous multi-sensor error analysis device provided in an embodiment of the present invention. Detailed Implementation

[0066] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0067] In a first aspect, embodiments of the present invention provide a spatiotemporal synchronized multi-sensor error analysis method, such as... Figure 1 As shown, the method may include the following steps:

[0068] S1: Acquire raw data collected by pre-deployed IoT sensors on the roadside from passing vehicles and perform preprocessing.

[0069] This invention allows for the pre-deployment of multiple IoT sensors on the same side of a road. During spatiotemporal synchronous multi-sensor error analysis, the data collected by the IoT sensors is directly acquired and processed. These IoT sensors include geomagnetic sensors and millimeter-wave radar; the geomagnetic sensor detects changes in the magnetic field as vehicles pass by, while the millimeter-wave radar acquires information such as the vehicle's position and speed.

[0070] In one alternative implementation, the process of pre-deploying IoT sensors on the roadside includes:

[0071] Multiple geomagnetic sensors are buried at a first preset interval on the road to the right of the vehicle's direction of travel;

[0072] Multiple millimeter-wave radars are installed at a second preset interval on the road to the right of the vehicle's direction of travel; wherein the receiving surface of the millimeter-wave radar faces the vehicle's direction of travel; and the height of the millimeter-wave radar above the ground is within a preset height range.

[0073] Please refer to the location of the IoT sensors. Figure 2 Understanding the diagram.

[0074] In one optional implementation, the geomagnetic sensor may be an RM3100 series triaxial geomagnetic sensor, but is not limited to other geomagnetic sensors on the market that can detect magnetic field disturbances caused by vehicles; the millimeter-wave radar may be a Huiershi DTAMD39-V, but is not limited to other millimeter-wave radars on the market that comply with regulations.

[0075] In this embodiment of the invention, the distance between adjacent geomagnetic sensors is a first preset spacing, which can be set as needed, for example, 15 meters. The distance between adjacent millimeter-wave radars is a second preset spacing, which can also be set as needed, for example, 200 meters.

[0076] In a preferred embodiment, the preset height range is 4m to 6m.

[0077] Since the geomagnetic sensor is buried in the ground and the millimeter-wave radar is erected at a certain height above the ground, in order to reduce the measurement error in the horizontal direction, in a preferred embodiment, the geomagnetic sensor and the millimeter-wave radar are placed parallel to each other.

[0078] Regarding S1, the acquisition and preprocessing of raw data collected by pre-deployed roadside IoT sensors from passing vehicles includes:

[0079] S11, acquire the disturbance signal carrying a timestamp collected when the geomagnetic sensor detects the vehicle passing by, filter the disturbance signal, and select the timestamp of the peak of the disturbance signal as the time when the vehicle passes the geomagnetic sensor.

[0080] Specifically, a geomagnetic sensor is a sensor based on the principle of the Earth's magnetic field. It can monitor changes in the Earth's magnetic field in real time and collect geomagnetic signals at its location. When a vehicle passes by, it causes disturbances in the geomagnetic signals. The geomagnetic sensor collects and stores these disturbance signals and timestamps them.

[0081] Since the disturbance signals collected by the geomagnetic sensor may be affected by factors such as noise, filtering is required to extract the true vehicle passing signal. After filtering the disturbance signal, the timestamp of the signal peak (i.e., the timestamp of the largest magnetic field disturbance) is selected as the time when the vehicle passes the geomagnetic sensor. Therefore, for each geomagnetic sensor, the preprocessed disturbance signal represents the position of the passing vehicle with the position of the geomagnetic sensor and carries the timestamp of the vehicle passing the geomagnetic sensor, which is used as the time when the vehicle was at that position, thereby obtaining the spatiotemporal information of the vehicle passing.

[0082] S12, acquire point cloud data carrying timestamps obtained by converting the vehicle reflected signal received by the millimeter-wave radar, cluster the point cloud data to generate vehicle trajectory data, and obtain the vehicle's position, speed and time.

[0083] Specifically, millimeter-wave radar is a radar technology that uses millimeter-wave electromagnetic waves for detection, and can collect point cloud data within its detection range. When a vehicle passes through the detection range of the millimeter-wave radar, the vehicle generates reflected signals. These reflected signals are received by the millimeter-wave radar, converted into point cloud data, and saved, while also being timestamped.

[0084] For any millimeter-wave radar, vehicle trajectory data for each detected vehicle can be generated by clustering its point cloud data. This vehicle trajectory data reflects the vehicle's movement and driving status, including vehicle position, speed, and timestamps. The clustering algorithm can be either k-means or DBSCAN, and there are no specific restrictions.

[0085] For details on the detection principles of geomagnetic sensors and millimeter-wave radar, please refer to the relevant technical explanations; they will not be explained in detail here.

[0086] S2, transform the preprocessed data from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data;

[0087] It is understandable that the data collected and processed by geomagnetic sensors and millimeter-wave radar are based on their own sensor coordinate systems. In order to achieve data consistency and accuracy, a unified coordinate system is required when performing multi-source sensor data association and data fusion.

[0088] Reference Figure 2Further explanations are provided regarding the coordinate systems of various IoT sensors and the global coordinate system.

[0089] In one optional implementation, S2 may include the following steps:

[0090] S21. Based on the actual road and the installation location of each IoT sensor, select a reference coordinate system as a unified global coordinate system, and determine the position of each IoT sensor in the global coordinate system.

[0091] First, in this embodiment of the invention, a reference coordinate system is selected as a unified global coordinate system based on the actual road conditions and the installation locations of each IoT sensor. This global coordinate system is a fixed Cartesian coordinate system that accurately reflects the actual road conditions and marks the locations of each sensor within the global coordinate system.

[0092] S22, Based on the position of each IoT sensor in the global coordinate system, establish the corresponding coordinate transformation relationship from the IoT sensor coordinate system to the global coordinate system;

[0093] Understandably, because each IoT sensor is located in a different place, the coordinate transformation relationship between the IoT sensor coordinate system and the global coordinate system will also be different.

[0094] Taking a millimeter-wave radar as an example, coordinate transformation should be achieved in the following way:

[0095] Let the coordinates of the millimeter-wave radar measurement point in the millimeter-wave radar sensor coordinate system be (x...). r ,y r By first rotating the millimeter-wave radar sensor coordinate system counterclockwise by α and then translating it along the positive y-axis by d, the global coordinate system is obtained, thus representing the millimeter-wave radar measurement point in the global coordinate system (x). g ,y g ), introducing intermediate variables Then we have:

[0096]

[0097] The aforementioned transformation parameters α and d are preset based on the millimeter-wave radar. The coordinate transformation relationships of different IoT sensors can be understood by referring to the above process and combining it with the existing coordinate system transformation process. However, it should be noted that the transformation parameters corresponding to different IoT sensors are set separately and are not the same.

[0098] S23. For each IoT sensor, based on the coordinate transformation relationship from its own IoT sensor coordinate system to the global coordinate system, the preprocessed data of the sensor is transformed to the global coordinate system to obtain the coordinate transformed data corresponding to the IoT sensor.

[0099] It is understandable that each IoT sensor can use its own coordinate transformation relationship to transform its preprocessed data into the global coordinate system. The specific process will not be described in detail here.

[0100] S3, Establish a spatiotemporal synchronization error model;

[0101] Spatiotemporal synchronization error manifests as discrepancies in the measurements taken by various IoT sensors of the same vehicle at the same location at the same time. Therefore, in this embodiment of the invention, the spatiotemporal synchronization error model is used to characterize the difference in the location acquired by the geomagnetic sensor and the millimeter-wave radar of the same vehicle at the same time; the spatiotemporal synchronization error model consists of three parts: time synchronization error, spatial synchronization error, and sensor measurement error.

[0102] In one optional implementation, S3 may include the following steps:

[0103] S31, based on the time synchronization error caused by the asynchrony between the geomagnetic sensor time axis and the millimeter-wave radar time axis, and the vehicle speed, the expression for the time synchronization error is determined as follows:

[0104] δ(t)=v*dt (2);

[0105] Where v is the vehicle speed; d(t) is the time synchronization error, d(t) = t b -t r ;t b and t r These are the timestamps from the geomagnetic sensor and the millimeter-wave radar in the coordinate-transformed data, respectively, for the same location and the same vehicle.

[0106] The speed of each vehicle can be obtained from the coordinate-transformed data.

[0107] S32, Based on the difference between the installation error of the geomagnetic sensor and the installation error of the millimeter-wave radar, the expression for the space synchronization error is determined as follows:

[0108] δ(s)=ds b -ds r (3);

[0109] ds b =s b -s' b (4);

[0110] ds r =s r -s' r (5);

[0111] Spatial synchronization error manifests as installation errors in the geomagnetic sensor and millimeter-wave radar; here, only the error in the vehicle's driving direction is considered. (Refer to...) Figure 3 The spatial synchronization error δ(s) of the spatiotemporal synchronization error model is further explained. b s represents the difference between the theoretical and actual installation locations of the geomagnetic sensor. b This represents the theoretical installation location of the geomagnetic sensor; s' b This refers to the actual installation location of the geomagnetic sensor; ds r s represents the difference between the theoretical and actual installation locations of the millimeter-wave radar; r This refers to the theoretical installation location of the millimeter-wave radar; s' r This refers to the actual installation location of the millimeter-wave radar; the theoretical installation location is its position in the global coordinate system.

[0112] S33, based on the fact that the actual measurement point measured by the millimeter-wave radar is the closest point on the vehicle to the millimeter-wave radar rather than the center point of the vehicle, and based on the possibility that the center point of the vehicle and the geomagnetic sensor are not aligned during the peak time of the disturbance signal measured by the geomagnetic sensor, the distribution characteristics of the sensor measurement error are determined to be: following a Gaussian distribution with a mean of half the length of the vehicle body.

[0113] Reference Figure 4 Further explanation is provided regarding the sensor measurement error δ(m) in the spatiotemporal synchronization error model. Taking the vehicle's midpoint as a reference point, the millimeter-wave radar measurement error manifests in the fact that the actual radar measurement point is not the vehicle's midpoint. Based on the millimeter-wave radar measurement principle and analysis of the original point cloud data, it should be the point closest to the radar. The geomagnetic sensor measurement error manifests in the fact that the vehicle's midpoint is not necessarily aligned with the geomagnetic sensor during the peak time of geomagnetic disturbances. This can be approximated as following a Gaussian distribution with the vehicle's midpoint as the mean. The sensor measurement error δ(m) is as follows... Figure 4 As shown.

[0114] S34, based on the time synchronization error, the spatial synchronization error, and the sensor measurement error, the expression for the spatiotemporal synchronization error model is established as follows:

[0115] d s =δ(t)+δ(s)+δ(m) (6);

[0116] Where δ(t) is the time synchronization error; δ(s) is the spatial synchronization error; and δ(m) is the sensor measurement error.

[0117] S4. Based on the coordinate transformation data obtained from all IoT sensors, the measurement error of the sensors is solved separately according to different vehicle models;

[0118] In one optional implementation, S4 may include the following steps:

[0119] S41, based on the distribution characteristics of the sensor measurement error, the coordinate transformation data obtained from all IoT sensors are divided into three types according to the vehicle body length: large vehicles, medium vehicles, and small vehicles. The general expression for the vehicle type of the sensor measurement error is determined as follows:

[0120] δ(m)=A*δ(m) big +B*δ(m) middle +C*δ(m) small (7);

[0121] The classification of large, medium, and small vehicles can be based on existing vehicle body lengths, divided into three non-overlapping ranges from largest to smallest. The specific range can be set as needed. For example, large vehicles are defined as those with a body length greater than 6 meters, medium vehicles as those with a body length between 4.25 meters and 6 meters, and small vehicles as those with a body length between 3.75 meters and 4.25 meters; δ(m) big For large vehicles, the sensor measurement error is δ(m). middle The sensor measurement error for a medium-sized vehicle; δ(m) small The sensor measurement error is for a small car; A, B, and C represent δ(m) respectively. big δ(m) middle δ(m) small The corresponding proportional coefficients; if the vehicle is a large vehicle, A=1, B=0, C=0 in the general expression for vehicle type; if the vehicle is a medium-sized vehicle, A=0, B=1, C=0 in the general expression for vehicle type; if the vehicle is a small vehicle, A=0, B=0, C=1 in the general expression for vehicle type.

[0122] Therefore, combining δ(t)=v*dt, d(t=t b -t r and d s =δ(t)+δ(s)+δ(m), d can be s Represented as:

[0123] d s = v*dt+δ(s)+A*δ(m) big +B*δ(m) middle +C*δ(m) small (8);

[0124] S42, based on the expression of the spatiotemporal synchronization error model, construct the calculation expressions for the spatiotemporal synchronization error models corresponding to large vehicles, medium-sized vehicles, and small vehicles respectively:

[0125] d s = v'*dt+δ(s)+δ(m) big(9);

[0126] d” s =v”*dt+δ(s)+δ(m) middle (10);

[0127] d”' s =v”'*dt+δ(s)+δ(m) small (11);

[0128] For formula (8), when the vehicles are large, medium, and small, formulas (9), (10), and (11) are obtained respectively. For all vehicles, dt and δ(s) are the same; v', v” and v”’ are the speeds of large, medium, and small vehicles respectively; d' s ,d” s and d”' s The spatiotemporal synchronization error data for large vehicles, medium vehicles, and small vehicles can be obtained by subtracting the position data of the geomagnetic sensor and millimeter-wave radar that perform position detection on the same vehicle at the same time from the coordinate-transformed data.

[0129] S43, calculate the difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and medium-sized vehicles to obtain a first expression; calculate the difference between the spatiotemporal synchronization error model calculation expressions corresponding to medium-sized vehicles and small vehicles to obtain a second expression; calculate the difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and small vehicles to obtain a third expression; wherein, the first expression, the second expression, and the third expression are respectively:

[0130] d' s -d” s = (v'-v”)*dt+(δ(m) big -δ(m) middle (12);

[0131] d” s -d”' s = (v”-v”')*dt+(δ(m) middle -δ(m) small (13);

[0132] d' s -d”' s = (v'-v”')*dt+(δ(m) big -δ(m) small (14);

[0133] S44: Obtain the spatiotemporal synchronization error data and speed of all large, medium and small vehicles from the coordinate transformation data obtained from all IoT sensors, and merge them into a dataset;

[0134] S45, based on the data of each vehicle type in the dataset, fit and solve the first expression, the second expression, and the third expression to obtain the sensor measurement error and time synchronization error corresponding to large vehicles, medium vehicles, and small vehicles, respectively.

[0135] Specifically, this invention takes into account that the length of a small car is typically between 3.75m and 4.25m, and sets δ(m) as an example. small =2m. For formulas (12) to (14), substitute the data d for different vehicle models. s Using v, we can fit and solve for the time synchronization error dt and the measurement error δ(m) for each vehicle model. big δ(m) middle δ(m) small The difference between them δ(m) big -δ(m) middle , δ(m) big -δ(m) small , δ(m) middle -δ(m) small .

[0136] Based on the measurement characteristics of millimeter-wave radar and geomagnetic sensors, the model can be considered to follow a Gaussian distribution with a mean of half the vehicle length. Therefore, the model can be fitted using the least squares method. With sufficient sample data, the measurement points will be evenly distributed on both sides of the fitting function. Thus, based on the number of data points required for least squares fitting, different spatiotemporal synchronization error data and speeds for each vehicle model can be obtained from the dataset. The fitting and solving formulas (12), (13), and (14) yield the common δ(m) for all large vehicles. big And dt, δ(m) is common to all mid-sized cars. middle And dt, and δ(m) common to all small cars. small And dt, where the common δ(m) of small cars small It's preset; the DT (data point) is actually the same for all car models.

[0137] The fitting process mainly includes: organizing the sample data and fitting it using the polyfit function in MATLAB. Refer to formulas (12) to (14) to calculate the difference in δ(m), i.e., δ(m). big -δ(m) middle , δ(m) big -δ(m) small , δ(m) middle -δ(m)small As the first parameter; the difference between ds, i.e., d' s -d” s ,d” s -d”' s ,d' s -d”' s As the second parameter, and with the third parameter of the function set to 1, a linear function fitting is performed. This process yields the time synchronization error dt and the sensor measurement errors for different vehicle models.

[0138] S5, based on the sensor measurement errors obtained from different vehicle models, obtains the spatial synchronization error of different vehicle models through fitting.

[0139] In one optional implementation, S5 may include:

[0140] For each vehicle model, based on the sensor measurement error and time synchronization error obtained under that model, as well as the spatiotemporal synchronization error data and speed of each vehicle under that model, the spatial synchronization error of that model is obtained by fitting and solving the spatiotemporal synchronization error model calculation expression.

[0141] The fitting solution can be achieved using the least squares method.

[0142] Specifically, as mentioned earlier, through S4, all large vehicles obtain a common δ(m). big And dt, all medium-sized cars obtain a common δ(m). middle And dt, all small cars obtain a common δ(m). small And dt, following a similar fitting solution method, the large vehicle utilizes the obtained δ(m) big And dt, as well as the spatiotemporal synchronization error data d' of specific vehicles under the large vehicle type. s Using the speed v', the spatiotemporal synchronization error model calculation expression corresponding to formula (9) is fitted and solved to obtain the common spatial synchronization error of large vehicles; medium-sized vehicles utilize the obtained δ(m) middle And dt, as well as the spatiotemporal synchronization error data d for specific vehicles under the medium model. s The spatial synchronization error is obtained by fitting and solving the spatiotemporal synchronization error model calculation expression corresponding to formula (11) with the speed v”, and the common spatial synchronization error of medium-sized vehicles is obtained; small vehicles use δ(m) small And dt, and the spatiotemporal synchronization error data d”' for specific vehicles under the small vehicle type. s And speed v”', fit and solve the spatiotemporal synchronization error model calculation expression corresponding to formula (12) to obtain the common spatial synchronization error of small car models;

[0143] The common time synchronization error of each model, the sensor measurement error corresponding to each model, and the spatial synchronization error can be used for subsequent parameter compensation to correct the spatiotemporal synchronization error.

[0144] Specifically, the common time synchronization error across all vehicle models is used as a parameter to compensate for the overall system's time axis. For the sensor measurement errors and spatial synchronization errors specific to each vehicle model, parameter compensation is applied separately for each vehicle model during subsequent data association. This method corrects the system's spatiotemporal synchronization error, resolves erroneous association issues caused by spatiotemporal synchronization problems, and optimizes subsequent data association.

[0145] This invention provides a spatiotemporal synchronization multi-sensor error analysis method to address the error problem existing in multi-sensor data fusion in intelligent transportation systems. By establishing a spatiotemporal synchronization model among multiple sensors, this invention performs spatiotemporal synchronization processing on the data from multiple sensors, solving the time and space alignment problem between different sensors. This improves the accuracy and reliability of the data, achieves precise calibration and optimization of sensor data, and resolves the problem of correlation errors caused by spatiotemporal synchronization errors in data from different sensor devices.

[0146] Secondly, corresponding to the above method embodiments, this invention also provides a spatiotemporal synchronized multi-sensor error analysis method apparatus, such as... Figure 5 As shown, the device includes:

[0147] The vehicle data acquisition and preprocessing module is used to acquire and preprocess raw data collected by roadside IoT sensors from passing vehicles; wherein, the IoT sensors include geomagnetic sensors and millimeter-wave radar.

[0148] The coordinate transformation module is used to transform the preprocessed data from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data;

[0149] A spatiotemporal synchronization error model establishment module is used to establish a spatiotemporal synchronization error model; wherein, the spatiotemporal synchronization error model is used to characterize the difference in position acquired by the geomagnetic sensor and the millimeter-wave radar of the same vehicle at the same moment; the spatiotemporal synchronization error model consists of three parts: time synchronization error, spatial synchronization error, and sensor measurement error;

[0150] The sensor measurement error solving module is used to solve the sensor measurement error separately according to different vehicle models based on the coordinate transformation data obtained from all IoT sensors.

[0151] The spatial synchronization error solving module is used to obtain the spatial synchronization error of different vehicle models by fitting the sensor measurement errors obtained from different vehicle models.

[0152] For details on the specific processing procedures of each module of the device, please refer to the relevant content in the first section, which will not be repeated here.

[0153] This invention provides a spatiotemporal synchronized multi-sensor error analysis device to address the error problem existing in multi-sensor data fusion in intelligent transportation systems. By establishing a spatiotemporal synchronization model among multiple sensors, this invention performs spatiotemporal synchronization processing on the data from multiple sensors to obtain sensor measurement errors and spatial synchronization errors corresponding to each vehicle model. These errors are then used for subsequent parameter compensation, correcting the system's spatiotemporal synchronization error and resolving the time and space alignment problem between different sensors. This improves the accuracy and reliability of the data, achieves precise calibration and optimization of sensor data, and solves the problem of association errors caused by spatiotemporal synchronization errors in data from different sensor devices, thus optimizing subsequent data association.

[0154] It should be noted that in the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0155] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A spatiotemporal synchronized multi-sensor error analysis method, characterized in that, include: The system acquires raw data collected by roadside IoT sensors from passing vehicles and performs preprocessing; the IoT sensors include geomagnetic sensors and millimeter-wave radar. The preprocessed data is transformed from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data. A spatiotemporal synchronization error model is established; wherein, the spatiotemporal synchronization error model is used to characterize the difference in position acquired by the geomagnetic sensor and the millimeter-wave radar of the same vehicle at the same moment; the spatiotemporal synchronization error model consists of three parts: time synchronization error, spatial synchronization error, and sensor measurement error; Based on the coordinate transformation data obtained from all IoT sensors, the measurement errors of the sensors are solved separately for different vehicle models; Based on the sensor measurement errors obtained from different vehicle models, the spatial synchronization errors of different vehicle models are obtained through fitting. The establishment of the spatiotemporal synchronization error model includes: Based on the time synchronization error caused by the asynchrony between the geomagnetic sensor time axis and the millimeter-wave radar time axis, and the vehicle speed, the expression for the time synchronization error is determined as follows: ； in, The speed of the vehicle; The time synchronization error is... ; and These are the timestamps from the geomagnetic sensor and the millimeter-wave radar in the coordinate-transformed data, respectively, for the same location and the same vehicle. Based on the difference between the installation error of the geomagnetic sensor and the installation error of the millimeter-wave radar, the expression for the space synchronization error is determined as follows: ； ； ； in, This represents the difference between the theoretical installation location and the actual installation location of the geomagnetic sensor. This is the theoretical installation location for the geomagnetic sensor; This refers to the actual installation location of the geomagnetic sensor. This represents the difference between the theoretical installation location and the actual installation location of the millimeter-wave radar. This represents the theoretical installation location for the millimeter-wave radar. This refers to the actual installation location of the millimeter-wave radar. Based on the fact that the actual measurement point measured by the millimeter-wave radar is the closest point on the vehicle to the millimeter-wave radar rather than the center point of the vehicle, and based on the possibility that the center point of the vehicle and the geomagnetic sensor are not aligned during the peak time of the disturbance signal measured by the geomagnetic sensor, the distribution characteristics of the sensor measurement error are determined to be: following a Gaussian distribution with a mean of half the length of the vehicle body. Based on the time synchronization error, the spatial synchronization error, and the sensor measurement error, the expression for the spatiotemporal synchronization error model is established as follows: ； in, This is due to time synchronization error; This refers to spatial synchronization error; For sensor measurement error; The process of calculating the sensor measurement errors based on the coordinate transformation data obtained from all IoT sensors, according to different vehicle models, includes: Based on the distribution characteristics of the sensor measurement errors, the coordinate transformation data obtained from all IoT sensors are categorized into three types according to vehicle length: large vehicles, medium vehicles, and small vehicles. The general expression for the vehicle type of the sensor measurement errors is determined as follows: ； in, For sensor measurement errors in large vehicles; For sensor measurement errors in mid-sized vehicles; For sensor measurement errors in small cars; , , They are respectively , , The corresponding proportional coefficient; if the vehicle is a large vehicle, the general expression for the vehicle type is as follows: , , If the vehicle is a mid-size vehicle, the general expression for the vehicle type is as follows: , , If the vehicle is a small car, the general expression for the vehicle type is as follows: , , ; Based on the expression of the spatiotemporal synchronization error model, the calculation expressions for the spatiotemporal synchronization error models corresponding to large vehicles, medium-sized vehicles, and small vehicles are constructed respectively as follows: ； ； ； in, , and The speeds of large vehicles, medium vehicles, and small vehicles are respectively; The difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and medium-sized vehicles is obtained to get the first expression; the difference between the spatiotemporal synchronization error model calculation expressions corresponding to medium-sized vehicles and small vehicles is obtained to get the second expression; the difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and small vehicles is obtained to get the third expression; wherein, the first expression, the second expression, and the third expression are respectively: ； ； ； The spatiotemporal synchronization error data and speed of all large, medium and small vehicles are obtained from the coordinate transformation data of all IoT sensors and merged into a dataset. Based on the data of each vehicle type in the dataset, the first expression, the second expression, and the third expression are fitted and solved to obtain the sensor measurement error and time synchronization error corresponding to large vehicles, medium vehicles, and small vehicles, respectively.

2. The spatiotemporal synchronized multi-sensor error analysis method according to claim 1, characterized in that, The process of pre-deploying IoT sensors on the roadside includes: Multiple geomagnetic sensors are buried at a first preset interval on the road to the right of the vehicle's direction of travel; Multiple millimeter-wave radars are installed at a second preset interval on the road to the right of the vehicle's direction of travel; wherein the receiving surface of the millimeter-wave radar faces the vehicle's direction of travel; and the height of the millimeter-wave radar above the ground is within a preset height range.

3. The spatiotemporal synchronized multi-sensor error analysis method according to claim 2, characterized in that, The preset height range is 4m to 6m.

4. The spatiotemporal synchronized multi-sensor error analysis method according to claim 1, characterized in that, The process of acquiring raw data collected by roadside IoT sensors from passing vehicles and preprocessing it includes: The disturbance signal carrying a timestamp is collected when the geomagnetic sensor detects the passing of a vehicle. After filtering the disturbance signal, the timestamp of the peak of the disturbance signal is selected as the time when the vehicle passes the geomagnetic sensor. The system acquires point cloud data with timestamps obtained by converting the reflected signals from the vehicle received by the millimeter-wave radar. It then clusters the point cloud data to generate vehicle trajectory data, thereby obtaining the vehicle's position, speed, and time.

5. The spatiotemporal synchronized multi-sensor error analysis method according to claim 4, characterized in that, The process of transforming the preprocessed data from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data includes: Based on the actual road conditions and the installation locations of each IoT sensor, a reference coordinate system is selected as a unified global coordinate system, and the position of each IoT sensor in the global coordinate system is determined. Based on the position of each IoT sensor in the global coordinate system, establish the corresponding coordinate transformation relationship from the IoT sensor coordinate system to the global coordinate system; For each IoT sensor, based on the coordinate transformation relationship from its own IoT sensor coordinate system to the global coordinate system, its preprocessed data is transformed to the global coordinate system to obtain the coordinate-transformed data corresponding to that IoT sensor.

6. The spatiotemporal synchronized multi-sensor error analysis method according to claim 1, characterized in that, The sensor measurement errors obtained based on different vehicle models are used to calculate the spatial synchronization errors of different vehicle models through a fitting method, including: For each vehicle model, based on the sensor measurement error and time synchronization error obtained under that model, as well as the spatiotemporal synchronization error data and speed of each vehicle under that model, the spatial synchronization error of that model is obtained by fitting and solving the spatiotemporal synchronization error model calculation expression.

7. The spatiotemporal synchronized multi-sensor error analysis method according to claim 1, characterized in that, The fitting solution is obtained using the least squares method.

8. A spatiotemporal synchronized multi-sensor error analysis device, characterized in that, include: The vehicle data acquisition and preprocessing module is used to acquire and preprocess raw data collected by roadside IoT sensors from passing vehicles; wherein, the IoT sensors include geomagnetic sensors and millimeter-wave radar. The coordinate transformation module is used to transform the preprocessed data from the IoT sensor coordinate system to the global coordinate system to obtain the coordinate-transformed data; A spatiotemporal synchronization error model establishment module is used to establish a spatiotemporal synchronization error model; wherein, the spatiotemporal synchronization error model is used to characterize the difference in position acquired by the geomagnetic sensor and the millimeter-wave radar of the same vehicle at the same moment; the spatiotemporal synchronization error model consists of three parts: time synchronization error, spatial synchronization error, and sensor measurement error; The sensor measurement error solving module is used to solve the sensor measurement error separately according to different vehicle models based on the coordinate transformation data obtained from all IoT sensors. The spatial synchronization error solving module is used to obtain the spatial synchronization error of different vehicle models by fitting the sensor measurement errors obtained from different vehicle models. The spatiotemporal synchronization error model establishment module establishes a spatiotemporal synchronization error model, specifically for: Based on the time synchronization error caused by the asynchrony between the geomagnetic sensor time axis and the millimeter-wave radar time axis, and the vehicle speed, the expression for the time synchronization error is determined as follows: ； in, The speed of the vehicle; The time synchronization error is... ; and These are the timestamps from the geomagnetic sensor and the millimeter-wave radar in the coordinate-transformed data, respectively, for the same location and the same vehicle. Based on the difference between the installation error of the geomagnetic sensor and the installation error of the millimeter-wave radar, the expression for the space synchronization error is determined as follows: ； ； ； in, This represents the difference between the theoretical installation location and the actual installation location of the geomagnetic sensor. This is the theoretical installation location for the geomagnetic sensor; This refers to the actual installation location of the geomagnetic sensor. This represents the difference between the theoretical installation location and the actual installation location of the millimeter-wave radar. This represents the theoretical installation location for the millimeter-wave radar. This refers to the actual installation location of the millimeter-wave radar. Based on the fact that the actual measurement point measured by the millimeter-wave radar is the closest point on the vehicle to the millimeter-wave radar rather than the center point of the vehicle, and based on the possibility that the center point of the vehicle and the geomagnetic sensor are not aligned during the peak time of the disturbance signal measured by the geomagnetic sensor, the distribution characteristics of the sensor measurement error are determined to be: following a Gaussian distribution with a mean of half the length of the vehicle body. Based on the time synchronization error, the spatial synchronization error, and the sensor measurement error, the expression for the spatiotemporal synchronization error model is established as follows: ； in, This is due to time synchronization error; This refers to spatial synchronization error; For sensor measurement error; The sensor measurement error solving module, based on the coordinate-transformed data obtained from all IoT sensors, solves for the sensor measurement errors separately for different vehicle models. Specifically, it is used for: Based on the distribution characteristics of the sensor measurement errors, the coordinate transformation data obtained from all IoT sensors are categorized into three types according to vehicle length: large vehicles, medium vehicles, and small vehicles. The general expression for the vehicle type of the sensor measurement errors is determined as follows: ； in, For sensor measurement errors in large vehicles; For sensor measurement errors in mid-sized vehicles; For sensor measurement errors in small cars; , , They are respectively , , The corresponding proportional coefficient; if the vehicle is a large vehicle, the general expression for the vehicle type is as follows: , , If the vehicle is a mid-size vehicle, the general expression for the vehicle type is as follows: , , If the vehicle is a small car, the general expression for the vehicle type is as follows: , , ; Based on the expression of the spatiotemporal synchronization error model, the calculation expressions for the spatiotemporal synchronization error models corresponding to large vehicles, medium-sized vehicles, and small vehicles are constructed respectively as follows: ； ； ； in, , and The speeds of large vehicles, medium vehicles, and small vehicles are respectively; The difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and medium-sized vehicles is obtained to get the first expression; the difference between the spatiotemporal synchronization error model calculation expressions corresponding to medium-sized vehicles and small vehicles is obtained to get the second expression; the difference between the spatiotemporal synchronization error model calculation expressions corresponding to large vehicles and small vehicles is obtained to get the third expression; wherein, the first expression, the second expression, and the third expression are respectively: ； ； ； The spatiotemporal synchronization error data and speed of all large, medium and small vehicles are obtained from the coordinate transformation data of all IoT sensors and merged into a dataset. Based on the data of each vehicle type in the dataset, the first expression, the second expression, and the third expression are fitted and solved to obtain the sensor measurement error and time synchronization error corresponding to large vehicles, medium vehicles, and small vehicles, respectively.

Images