Test vehicle fusion positioning method

Abstract

The invention provides a test vehicle fusion positioning method, and the method comprises the steps: obtaining a UWB positioning parameter vector and an AI visual positioning parameter vector of a test vehicle, forming a vector matrix, and carrying out the training of a positioning deep learning model based on a plurality of groups of vector matrixes; predicting the position coordinates of the test vehicle based on a positioning deep learning model; calculating the difference between the predicted position coordinate of the model and the RTK position coordinate of the tested vehicle under the normal signal; and adjusting model parameters of the positioning deep learning model based on the gap to obtain an optimized positioning deep learning model. According to the method, UWB positioning, AI visual positioning, deep learning model positioning, RTK positioning and other positioning methods are fused, UWB (ultra wide band) positioning data are taken as underlying data, an AI visual and deep learning model fusion technology is used, the problem of low positioning precision is solved, standard test positioning data is generated, and the effective rate and the passing rate of a test are improved.

Description

Technical Field

[0001] The present invention relates to the field of V2X testing, and more specifically, to a method for fusing the positioning of test vehicles. Background Art

[0002] In the current era of the booming development of intelligent transportation and autonomous driving, as a key enabling technology, V2X plays a crucial role in vehicle positioning and information interaction. To ensure the safety and reliability of V2X technology and autonomous driving systems, a large number of tests are carried out in closed test fields.

[0003] Closed test fields can simulate a rich variety of real-world scenarios, creating a controllable test environment for V2X devices and vehicles, thus facilitating a comprehensive evaluation and optimization of their performance. However, in relevant standards such as CSAE 246-2022, a high positioning accuracy requirement of up to two centimeters is imposed on V2X test vehicles in closed test fields. However, in reality, during the existing V2X tests in closed test fields, the positioning accuracy is often restricted by many factors and fails to meet the requirements of the test standards.

[0004] Satellite positioning signals, as an important basis for vehicle positioning, are easily interfered by the surrounding environment in practical applications. For example, "buildings" in closed test fields can block satellite signals, resulting in a significant decrease in the positioning accuracy based on satellite navigation. This not only seriously affects the accuracy of test results but also fails to meet the requirements of the standards for effectiveness and passability.

[0005] In closed test areas with special environments such as signal shielding chambers and under viaducts, although UWB positioning devices are used, their positioning accuracy still cannot meet the stringent requirements of the test standards. This greatly reduces the reliability of test data, fails to provide strong support for the optimization and improvement of V2X technology, and further hinders the development and application of intelligent transportation and autonomous driving technologies. Summary of the Invention

[0006] The present invention aims to solve the problem of insufficient vehicle positioning accuracy in weak satellite signal environments in existing closed test fields, and provides a method for fusing the positioning of test vehicles, including: Obtaining the UWB positioning parameter vector and the AI vision positioning parameter vector of the test vehicle, where the UWB positioning parameter vector is expressed as where, is the moment when the test vehicle is powered on to start positioning, is the time required for the j-th processing stage of the i-th base station, and the AI vision positioning parameter vector is expressed as where t is the translation vector of the camera, , The abscissa and ordinate of the position coordinates of the AI visual positioning of the vehicle to be tested respectively; Form a vector matrix with the UWB positioning parameter vector and the AI visual positioning parameter vector, and obtain multiple groups of vector matrices as multiple training samples; Train the positioning deep learning model based on multiple training samples; Predict the position coordinates of the test vehicle based on the trained positioning deep learning model; Calculate the gap between the predicted position coordinates of the test vehicle by the positioning deep learning model and the RTK position coordinates of the test vehicle under normal signals; Adjust the model parameters of the positioning deep learning model based on the gap to obtain the optimized positioning deep learning model; Predict the position coordinates of the vehicle to be tested based on the optimized positioning deep learning model.

[0007] A test vehicle fusion positioning method provided by the present invention fuses multiple positioning methods such as UWB positioning, AI visual positioning, deep learning model positioning, and RTK positioning, realizes UWB (ultra-wideband) positioning data as the underlying data, uses AI vision and deep learning model fusion technology to make up for the problem of low positioning accuracy, generates standard test positioning data, and improves the efficiency and passing rate of the test. Description of the Drawings

[0008] Figure 1 It is a flowchart of a test vehicle fusion positioning method provided by the present invention. Detailed Embodiments

[0009] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. In addition, the technical features in each embodiment or a single embodiment provided by the present invention can be combined with each other arbitrarily to form a feasible technical solution. This combination is not restricted by the order of steps and / or the pattern of structural composition, but must be based on what can be achieved by those of ordinary skill in the art. When the combination of technical solutions is contradictory or cannot be realized, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0010] Currently, although the closed test field provides a controllable test environment for V2X devices and vehicles, when facing complex environments such as signal shielding warehouses and under viaducts, the existing positioning technologies are difficult to achieve the ideal accuracy requirements. Since satellite signals are easily blocked by surrounding buildings, the positioning method based on satellite navigation often shows a decrease in accuracy. Although UWB positioning devices are used as a supplement, the effect is still not satisfactory and cannot meet the strict requirements for a positioning accuracy of two centimeters in relevant standards.

[0011] To overcome these problems, the present invention has made innovations and improvements in multiple aspects. First, by integrating multiple positioning technologies, it no longer simply relies on satellite navigation or single UWB positioning, but comprehensively utilizes the advantages of various technologies such as AI vision and deep learning to achieve the fusion and complementarity of multi-source data. In addition, after entering the normal signal area, the prediction data is corrected according to the normal RTK signal docking data to further improve the positioning accuracy.

[0012] See Figure 1 , the present invention provides a method for fusing positioning of a test vehicle, and the method includes: Step 1, obtaining the UWB positioning parameter vector and the AI vision positioning parameter vector of the test vehicle, where the UWB positioning parameter vector is expressed as , where is the moment when the test vehicle powers on for positioning, is the time required for the j-th processing stage of the i-th base station, and the AI vision positioning parameter vector is expressed as , where t is the translation vector of the camera, , are respectively the abscissa and ordinate of the position coordinates of the AI vision positioning of the test vehicle to be tested.

[0013] It can be understood that the embodiment of the present invention fuses the UWB positioning method and the AI vision positioning method, and then trains the deep learning model. The UWB positioning parameter vector and the AI vision positioning parameter vector of the test vehicle are respectively obtained, and subsequently, the positioning deep learning model is trained based on the UWB positioning parameter vector and the AI vision positioning parameter vector.

[0014] Among them, the principle of positioning the test vehicle based on the UWB positioning method is: The position coordinates of the test vehicle to be tested can be obtained by using the trilateration method, and at least 3 base stations are required. Let i be the base station serial number, then the signal transmission time from the i-th base station to the test vehicle is: ; where i represents the serial number of the base station, , , and respectively represent the time required for the i-th base station in 4 signal processing stages.

[0015] The distance from the i-th base station to the test vehicle is: ; where c is the pulse velocity.

[0016] Locating the test vehicle using the UWB positioning method is specifically as follows: According to the coordinates of 3 base stations and the coordinates of the test vehicle to be determined, calculate the distance between each base station and the test vehicle to be determined: ; where are the coordinates of the test vehicle to be determined, represents the coordinates of the i-th base station.

[0017] Combine: ; Solve for the coordinates of the test vehicle to be determined .

[0018] where, take as the UWB positioning parameter vector.

[0019] Locating the vehicle based on the AI vision positioning method, the specific process is as follows: Capture at a frequency of 100hz. First, perform camera calibration to calibrate the camera parameters, including the rotation matrix R and translation vector t in the world coordinate system of the camera, the focal length fx , fy , the principal point coordinates cx , cy and distortion coefficients, which are obtained through the calibration tool: .

[0020] Then use the Faster R-CNN algorithm to detect the vehicle and output the center point or corner pixel coordinates of the bounding box of the test vehicle ([[]] u , v ). Perform coordinate conversion. Let the altitude be , and the time be T. Convert the pixel coordinates to world coordinates through inverse perspective mapping (IPM), correct the image using the calibrated distortion coefficients, and obtain the mapping matrix H from the image coordinate system to the ground coordinate system through calibration. Convert the pixel coordinates ([[]] u , v ) to the ground coordinates : .

[0021] Add as a parameter to the AI visual positioning parameter vector.

[0022] Step 2: Combine the UWB positioning parameter vector and the AI visual positioning parameter vector to form a vector matrix, and obtain multiple groups of vector matrices as multiple training samples.

[0023] It can be understood that combining the UWB positioning parameter vector and the AI visual positioning parameter vector to form a vector matrix is expressed as follows: ; where the UWB parameter vector must be normalized and expressed as: .

[0024] Take each vector matrix as a training sample to obtain multiple training samples.

[0025] Step 3: Train the positioning deep learning model based on multiple training samples.

[0026] It can be understood that multiple training samples are obtained, and the positioning deep learning model is trained based on multiple training samples.

[0027] First, design the positioning deep learning model. In the embodiment of the present invention, the positioning deep learning model uses the Same convolution. The positioning deep learning model includes an input layer, a convolutional layer, a pooling layer, a Flatten layer, a fully connected layer, and an output layer. Training the positioning deep learning model based on multiple training samples includes: Input the vector matrix into the input layer, and perform convolution on the vector matrix based on the convolutional layer; Perform max pooling processing on the output result of the convolutional layer based on the pooling layer; Flatten the output matrix of the pooling layer into a flat one-dimensional vector based on the Flatten layer; Output the one-dimensional vector output by the Flatten layer as a two-dimensional vector based on the fully connected layer; Output the predicted position coordinates of the test vehicle through the output layer.

[0028] Among them, during the training of the positioning deep learning model, the loss value of the positioning deep learning model is calculated based on the fusion objective function. The fusion objective function is: ; where Loss represents the loss value of the positioning deep learning model, represents the predicted position coordinates of the test vehicle by the deep learning model at time k, represents the UWB positioning coordinates of the test vehicle at time k, It represents the AI visual positioning position coordinates of the test vehicle at time k, and n represents the number of time moments, which can also be understood as n training samples.

[0029] The positioning deep learning model is adjusted based on the loss value, for example, the network structure of the positioning deep learning model is adjusted to minimize the loss.

[0030] Step 4: predict the position coordinates of the test vehicle based on the trained positioning deep learning model.

[0031] It is understandable that after the positioning deep learning model is trained in step 3, the trained positioning deep learning model is used to test the test set, and the positioning deep learning model is optimized based on the test results of the test set.

[0032] Specifically, the UWB positioning parameter vector and the AI visual positioning parameter vector of the test vehicle are obtained as test samples according to the preset sampling time interval T5; m test samples with time conversion gaps are obtained, where m is a positive integer and can be understood as test samples at m moments; the m test samples are predicted based on the trained positioning deep learning model to obtain the predicted position coordinates of the m test samples, which are expressed as: .

[0033] Step 5: Calculate the difference between the predicted position coordinates of the test vehicle by the positioning deep learning model and the RTK position coordinates of the test vehicle under normal signals.

[0034] It is understandable that the RTK position coordinates of the test vehicle at the current moment are obtained after the test vehicle is in the normal signal area. , sample the RTK signal at the same preset sampling time interval T5, and test the real-time speed of the vehicle at the current moment and the angle between the test vehicle and the horizontal axis of the coordinate system , estimate the RTK position coordinates of the test vehicle at the previous moment ,in: ; ; in, is the RTK position coordinate of the test vehicle at the current time m, is the estimated RTK position coordinate of the test vehicle at the last moment m-1.

[0035] The RTK position coordinates of the test vehicle at m moments are estimated using the same estimation method. Calculate the gap between the predicted position coordinates of the m test samples of the test vehicle according to the positioning deep learning model and the estimated RTK position coordinates of the m test vehicles. The gap is calculated by the formula: ; where represents the RTK position coordinates at time i, represents the predicted position coordinates of the test vehicle at time g output by the positioning deep learning model.

[0036] Step 6: Based on the gap, adjust the model parameters of the positioning deep learning model to obtain the optimized positioning deep learning model.

[0037] It can be understood that based on the gap adjust the model parameters of the positioning deep learning model to obtain the optimized positioning deep learning model. Specifically, it includes: Set the step size adjustment parameter ω of the positioning deep learning model. Based on the step size adjustment parameter ω, update the fusion objective function of the positioning deep learning model: ; where represents the loss value of the positioning deep learning model, represents the predicted position coordinates of the test vehicle by the deep learning model at time k, represents the UWB positioning position coordinates of the test vehicle at time k, represents the AI vision positioning position coordinates of the test vehicle at time k, and n represents the number of time instances.

[0038] Continuously adjust the step size adjustment parameter ω of the positioning deep learning model to minimize the gap between the predicted position coordinates of the test vehicle by the positioning deep learning model and the RTK position coordinates under normal signals, and optimize the preliminarily trained positioning deep learning model to obtain the optimized positioning deep learning model.

[0039] Step 7: Based on the optimized positioning deep learning model, predict the position coordinates of the vehicle to be tested.

[0040] It can be understood that the above steps optimize the positioning deep learning model according to the RTK position coordinates of the test vehicle, and predict the position coordinates of the vehicle to be tested based on the optimized positioning deep learning model.

[0041] A method for testing vehicle integrated positioning provided by an embodiment of the present invention effectively integrates UWB positioning, AI vision positioning, a deep learning model, and RTK positioning. The positioning information provided by AI vision is optimized in the deep learning model and then complements the UWB positioning data. By means of a fusion algorithm, the positioning accuracy and reliability are improved. When satellite signals are blocked, AI vision can provide continuous positioning, overcoming the limitations of a single technology in complex environments. In addition, AI vision can accurately extract vehicle position data and establish a positioning model with centimeter-level accuracy, solving the problem of large positioning errors in a satellite signal-blocked environment in a closed test field, thereby improving the positioning accuracy, test efficiency, and passing rate.

[0042] It should be noted that in the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not described in detail in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0043] Persons skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0044] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0045] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0046] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the process Figure 1 one process or a plurality of processes and / or boxes Figure 1 steps for the functions specified in one box or a plurality of boxes.

[0047] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn the basic inventive concept. Therefore, the appended claims are intended to be construed to cover the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0048] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for testing vehicle integrated positioning, characterized in that, Including: Obtain the UWB positioning parameter vector and the AI vision positioning parameter vector of the test vehicle, and the UWB positioning parameter vector is expressed as , where is the moment when the test vehicle starts positioning after being powered on, is the time required for the j-th processing stage of the i-th base station, and the AI vision positioning parameter vector is expressed as , where t is the translation vector of the camera, , are the abscissa and ordinate of the position coordinates of the AI vision positioning of the test vehicle respectively; Combining the UWB positioning parameter vector and the AI vision positioning parameter vector to form a vector matrix, and obtaining multiple groups of vector matrices as multiple training samples; Training a positioning deep learning model based on multiple training samples; Predicting the position coordinates of a test vehicle based on the trained positioning deep learning model; Calculating the gap between the predicted position coordinates of the test vehicle by the positioning deep learning model and the RTK position coordinates of the test vehicle under normal signals; Adjusting the model parameters of the positioning deep learning model based on the gap to obtain the optimized positioning deep learning model; Predicting the position coordinates of a vehicle to be tested based on the optimized positioning deep learning model.

2. The test vehicle fusion positioning method according to claim 1, wherein Obtaining the AI vision positioning parameter vector includes: Calibrate the camera parameters, which include the rotation matrix of the camera in the world coordinate system R and the translation vector t、 the camera focal length fx and fy the principal point coordinates cx and cy as well as the distortion coefficients of the camera; Detect the test vehicle based on the Faster R-CNN algorithm and output the pixel position coordinates of the test vehicle ( u , v ); Based on the camera parameters, converting the pixel position coordinates of the test vehicle into ground coordinates: ； ； Among them, is the altitude of the test vehicle; Take as the AI vision positioning parameter vector.

3. The test vehicle integrated positioning method according to claim 1, wherein Combining the UWB positioning parameter vector and the AI vision positioning parameter vector to form a vector matrix, including: Normalize the UWB positioning parameter vector as follows: ； Combining the normalized UWB positioning parameter vector and the AI vision positioning parameter vector to form a vector matrix.

4. The test vehicle integrated positioning method according to claim 1, characterized in that, The positioning deep learning model includes an input layer, a convolutional layer, a pooling layer, a Flatten layer, a fully connected layer, and an output layer. Training the positioning deep learning model based on multiple training samples includes: Inputting the vector matrix into the input layer, and performing convolution on the vector matrix based on the convolutional layer; Performing max-pooling processing on the output result of the convolutional layer based on the pooling layer; Flattening the output matrix of the pooling layer into a flat one-dimensional vector based on the Flatten layer; Outputting the one-dimensional vector output by the Flatten layer as a two-dimensional vector based on the fully connected layer; Outputting the predicted position coordinates of the test vehicle through the output layer.

5. The test vehicle integrated positioning method according to claim 1 or 4, characterized in that When training the positioning deep learning model, calculating the loss value of the positioning deep learning model based on a fusion objective function, and the fusion objective function is: ； Among them, represents the loss value of the localization deep learning model, represents the predicted position coordinates of the test vehicle by the deep learning model at time k, represents the UWB localization position coordinates of the test vehicle at time k, represents the AI vision localization position coordinates of the test vehicle at time k, and n represents the number of time instances, that is, the number of training samples.

6. The test vehicle integrated positioning method according to claim 5, wherein, Obtaining the UWB positioning position coordinates of the test vehicle includes: Setting at least 3 base stations around the test vehicle, and the signal transmission time of each base station is: ； where \(i\) represents the serial number of the base station, 、 、 and respectively represent the time required for the \(i\)-th base station in four signal processing stages; The distance from the i-th base station to the test vehicle is: ； Where c is the pulse speed; Calculating the distance between each base station and the test vehicle to be determined according to the coordinates of 3 base stations and the coordinates of the test vehicle to be determined; ； Among them, is the coordinate of the test vehicle to be obtained, represents the coordinate of the i-th base station; Solve the joint equations to obtain the coordinates of the test vehicle to be determined .

7. The test vehicle integrated positioning method according to claim 1, wherein Obtaining the UWB positioning parameter vector and the AI vision positioning parameter vector of the test vehicle includes: Obtaining the UWB positioning parameter vector and the AI vision positioning parameter vector of the test vehicle at a preset sampling time interval T5 as test samples; Obtaining m test samples with a time conversion gap, where m is a positive integer; Predicting m test samples based on the trained positioning deep learning model to obtain the predicted position coordinates of the m test samples, denoted as: 。 8. The test vehicle integrated positioning method according to claim 7, wherein Calculating the gap between the predicted position coordinates of the test vehicle by the positioning deep learning model and the RTK position coordinates of the test vehicle under normal signals includes: Obtain the RTK position coordinates of the test vehicle at the current moment received after the test vehicle is in the normal signal area , according to the same preset sampling time interval T5, based on the real-time speed of the test vehicle at the current moment and the included angle between the test vehicle and the horizontal axis of the coordinate system , estimate the RTK position coordinates of the test vehicle at the previous moment , where: ； ； Among them, is the RTK position coordinate of the test vehicle at the current moment m, is the estimated RTK position coordinate of the test vehicle at the previous moment m - 1; Calculate the gap between the predicted position coordinates of the m test samples of the test vehicle according to the positioning deep learning model and the estimated RTK position coordinates of the m test vehicles.

9. The test vehicle fusion positioning method according to claim 8, wherein Calculating the difference between the predicted position coordinates of the positioning deep learning model and the RTK position coordinates , including: ； Among them, represents the RTK position coordinates at time i, represents the predicted position coordinates of the test vehicle at time g output by the positioning deep learning model.

10. The test vehicle integrated positioning method according to claim 9, wherein, adjusting the model parameters of the positioning deep learning model based on the gap to obtain the optimized positioning deep learning model, including: setting the step size adjustment parameter ω of the positioning deep learning model, and updating the fusion objective function of the positioning deep learning model based on the step size adjustment parameter ω: ； Among them, represents the loss value of the positioning deep learning model, represents the predicted position coordinates of the test vehicle by the deep learning model at time k, represents the UWB positioning position coordinates of the test vehicle at time k, represents the AI vision positioning position coordinates of the test vehicle at time k, and n represents the number of time instances.

Images