Vehicle trajectory positioning method, device and electronic equipment in a non-positioning signal scenario

Abstract

The application provides a vehicle trajectory positioning method, device and electronic equipment in a no-positioning-signal scenario. The method comprises: acquiring a multi-modal data stream of a vehicle in a target time period, the target time period being a time period from the vehicle driving out of the no-positioning-signal scenario to a positioning-signal scenario; performing a mirror flipping operation on the multi-modal data stream with the end timestamp of the multi-modal data stream as a center point to obtain an initial pseudo data stream of a preset time period of the vehicle; performing an inverse inversion processing on data values in the initial pseudo data stream to obtain a pseudo data stream, and inputting the pseudo data stream into a Kalman filter to perform fusion positioning processing on the pseudo data stream by the Kalman filter to obtain a driving trajectory of the vehicle in the target time period. The problem that the prior art needs to increase a reverse filter to realize vehicle positioning in a no-signal scenario by a forward and reverse filter and has a high no-signal positioning cost is solved.

Description

Technical Field

[0001] This application relates to the field of vehicle trajectory positioning technology in scenarios without positioning signals, and more specifically, to a vehicle trajectory positioning method, apparatus, computer-readable storage medium, and electronic device in scenarios without positioning signals. Background Technology

[0002] In high-precision autonomous driving and integrated navigation systems, GNSS / INS (Global Navigation Satellite System / Inertial Navigation System) fusion positioning technology is widely used in offline post-processing scenarios to generate high-precision trajectory ground truth values ​​to support tasks such as perception algorithm training, positioning system evaluation, and map building. However, in the early stages of system startup, due to high initial state uncertainty and insufficient sensor observation information, especially when the vehicle is in an environment with GNSS signal obstruction (such as underground parking garages or urban canyons), Kalman filters often face the core problems of initial alignment difficulties and slow convergence speed.

[0003] To solve this problem, existing technologies generally adopt the following two mainstream solutions:

[0004] Option 1 relies on long-term static initialization: This method requires the vehicle to remain absolutely stationary for several minutes to tens of minutes before starting. By accumulating the zero-bias estimation of the inertial measurement unit (IMU) and the stable convergence of the GNSS signal, the filter state is gradually aligned. While this method achieves convergence, it significantly increases data acquisition costs and testing cycles. Furthermore, it is difficult to maintain the static condition continuously in real-world applications (such as urban roads and multi-story parking garages), limiting its practicality. Additionally, in scenarios with signal obstruction, such as underground parking garages, even with prolonged static operation, the GNSS signal cannot achieve stable convergence.

[0005] Option 2 involves constructing a bidirectional extended Kalman filter (Backward EKF) architecture: In addition to the forward filter, an extra structurally symmetric, operationally reverse-biased inverse filter module is developed. This module compensates for forward initialization errors by back-calculating state estimates from the end of the trajectory to the starting point. While this method improves convergence accuracy, it suffers from significant technical intrusion: it requires maintaining independent state transition matrices, observation models, error propagation equations, and sensor fusion logic for both the forward and reverse systems. If the system adds new sensing modalities (such as vision, wheel speedometer, or lidar matching output) or updates the dynamic model, both filters must be reconstructed synchronously, leading to an exponential increase in system maintenance costs and introducing cross-module consistency errors, significantly reducing engineering robustness.

[0006] In summary, existing positioning filter initialization convergence schemes are inefficient when the vehicle is in an environment where GNSS signals are blocked. Scheme 1 relies on the long period of vehicle stationary operation and suffers from low efficiency. Scheme 2 requires the maintenance of forward and reverse filters, which has the problems of strong technical intrusion into the system, high cost, and serious waste of computing power. Summary of the Invention

[0007] The main objective of this application is to provide a vehicle trajectory localization method, device, computer-readable storage medium, and electronic device in a scenario without a positioning signal, so as to at least solve the problem that the existing technology of achieving vehicle localization in a scenario without a positioning signal requires the addition of a reverse filter, resulting in high cost for localization without a positioning signal.

[0008] To achieve the above objectives, according to one aspect of this application, a vehicle trajectory localization method is provided in a scenario without a positioning signal, comprising: acquiring a multimodal data stream of a vehicle within a target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives a positioning signal. The target time period is the time from when the vehicle leaves a scenario without a positioning signal to when it enters a scenario with a positioning signal. Using the end timestamp of the multimodal data stream as the center point, a mirror flip operation is performed on the multimodal data stream to obtain an initial pseudo-data stream for the vehicle within the preset time period. The beginning data of the initial pseudo-data stream is the end data of the multimodal data stream, and the end data of the initial pseudo-data stream is the beginning data of the multimodal data stream. The data values ​​in the initial pseudo-data stream are inverted to obtain a pseudo-data stream. This pseudo-data stream is used to initialize a Kalman filter, and the initialized Kalman filter is used to perform fusion positioning processing on the pseudo-data stream to obtain the driving trajectory corresponding to the preset time period. Then, according to the mirror mapping relationship of the mirror flip operation, the driving trajectory corresponding to the preset time period is restored using a time axis to obtain the driving trajectory of the vehicle within the target time period.

[0009] Optionally, the data values ​​in the initial pseudo data stream are inverted to obtain a pseudo data stream, including: identifying the first-order time-varying derivative in the initial pseudo data stream, wherein the first-order time-varying derivative includes at least one of the vehicle's RTK speed, chassis angular velocity, IMU angular velocity, and estimated gyroscope zero bias; and performing the inversion process on at least the first-order time-varying derivative in the initial pseudo data stream to obtain the pseudo data stream.

[0010] Optionally, at least the first-order time-varying derivative in the initial pseudo-data stream is inverted to obtain the pseudo-data stream, including: identifying the target gear information of the initial pseudo-data stream, wherein the target gear information includes at least one of the vehicle's parking gear, low gear, drive gear, and neutral gear; modifying the target gear information of the initial pseudo-data stream to reverse gear, and modifying the reverse gear in the initial pseudo-data stream to drive gear, and performing the inversion process on the first-order time-varying derivative in the initial pseudo-data stream to obtain the pseudo-data stream.

[0011] Optionally, after acquiring the multimodal data stream of the vehicle within the target time period, the method further includes: identifying the vehicle stationary periodic data stream in the multimodal data stream; and performing a trimming process on the vehicle stationary periodic data stream in the multimodal data stream to obtain a processed multimodal data stream.

[0012] Optionally, identifying the vehicle stationary periodic data stream in the multimodal data stream includes: identifying the vehicle stationary periodic data stream in the multimodal data stream by performing time-domain and frequency-domain detection on the IMU data in the multimodal data stream; and / or, identifying the vehicle stationary periodic data stream in the multimodal data stream by detecting the vehicle wheel speed data in the multimodal data stream.

[0013] Optionally, after obtaining the vehicle's driving trajectory within the target time period, the method further includes: evaluating and training the vehicle's localization algorithm using the driving trajectory, and evaluating and training the vehicle's perception algorithm.

[0014] Optionally, using the end timestamp of the multimodal data stream as the center point, a mirror flip operation is performed on the multimodal data stream to obtain the initial pseudo-data stream of the vehicle within a preset time period, including: using the formula: The multimodal data stream is mirrored and flipped to obtain the initial pseudo-data stream for the target time period of the vehicle, wherein... The center point, The original data timestamp of the multimodal data stream. The timestamp of the flipped data in the initial pseudo data stream.

[0015] According to another aspect of this application, a vehicle trajectory positioning device for scenarios without positioning signals is provided, comprising: an acquisition unit, configured to acquire a multimodal data stream of a vehicle within a target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives a positioning signal. The target time period is the time from when the vehicle leaves a scenario without a positioning signal to when it enters a scenario with a positioning signal. A mirror flip operation unit is used to perform a mirror flip operation on the multimodal data stream, centered on the end timestamp of the multimodal data stream, to obtain an initial pseudo-data stream for the vehicle within a preset time period. The beginning data of the initial pseudo-data stream is the end data of the multimodal data stream, and the end data of the initial pseudo-data stream is the beginning data of the multimodal data stream. A processing unit is used to invert the data values ​​in the initial pseudo-data stream to obtain a pseudo-data stream, initialize a Kalman filter using the pseudo-data stream, perform fusion positioning processing on the pseudo-data stream using the initialized Kalman filter to obtain the driving trajectory corresponding to the preset time period, and perform time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle within the target time period.

[0016] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to execute any of the vehicle trajectory positioning methods in the scenario without positioning signals described above.

[0017] According to another aspect of this application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any of the described vehicle trajectory localization methods in a scenario without a positioning signal.

[0018] By applying the technical solution of this application, after the vehicle regains its positioning signal, the historical multimodal data stream is mirrored and flipped with that moment as the time center. This generates a reverse pseudo-data stream that conforms to physical laws and is input into an unmodified forward Kalman filter, achieving "post-event vehicle trajectory reconstruction" without altering any underlying filtering architecture. This method completely avoids the highly invasive defect of traditional bidirectional filters requiring dual-system maintenance. By reconstructing the spatiotemporal structure of the data layer, the original filter can convert the reversed data into the true forward input, thus achieving zero errors and instantaneous convergence. This not only significantly reduces system upgrade costs but also, for the first time, achieves high-precision reconstruction of the complete trajectory of "no positioning signal sections" (such as underground parking garages) in offline post-processing scenarios. It fills the trajectory gap in the initialization period of traditional positioning systems, providing data integrity support for autonomous driving algorithm training, ground truth generation, and system evaluation. This solves the problem that existing technologies require the addition of a reverse filter to achieve vehicle positioning in no-signal scenarios through forward and reverse filters, resulting in high positioning costs in no-signal scenarios. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0020] Figure 1 A hardware structure block diagram of a mobile terminal for performing a vehicle trajectory positioning method in a scenario without a positioning signal, according to an embodiment of this application, is shown.

[0021] Figure 2 A flowchart illustrating a vehicle trajectory localization method in a scenario without a positioning signal, according to an embodiment of this application, is shown.

[0022] Figure 3 A flowchart illustrating a vehicle trajectory localization method in a specific scenario without a positioning signal, according to an embodiment of this application, is shown.

[0023] Figure 4 A structural block diagram of a vehicle trajectory positioning device in a scenario without a positioning signal, according to an embodiment of this application, is shown. Detailed Implementation

[0024] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] As described in the background section, existing technologies for vehicle positioning in no-signal scenarios require the addition of a reverse filter, resulting in high positioning costs. To address the issue of high positioning costs in existing technologies for vehicle positioning in no-signal scenarios, embodiments of this application provide a vehicle trajectory positioning method, apparatus, computer-readable storage medium, and electronic device in no-positioning-signal scenarios.

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0029] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a vehicle trajectory positioning method in a scenario without a positioning signal, according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0030] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the vehicle trajectory positioning method in the absence of a positioning signal in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0031] This embodiment provides a vehicle trajectory positioning method in a scenario without a positioning signal, which runs on a mobile terminal, computer terminal or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than that shown here.

[0032] Figure 2 This is a flowchart of a vehicle trajectory localization method in a scenario without a positioning signal, according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0033] Step S201: Obtain the multimodal data stream of the vehicle within the target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle leaves the scenario without a positioning signal to when it has a positioning signal.

[0034] The positioning signal is either the vehicle's GNSS positioning signal or a lidar-matched positioning signal.

[0035] Step S202: Using the end timestamp of the multimodal data stream as the center point, a mirror flip operation is performed on the multimodal data stream to obtain an initial pseudo-data stream for the vehicle's preset time period. The beginning data of the initial pseudo-data stream is the end data of the multimodal data stream, and vice versa. Specifically, the mirror flip operation is achieved by modifying the timestamps corresponding to the multimodal data. The target time period is the period from when the vehicle has no positioning signal to when it has a positioning signal. For example, if the target time period is 1:00 AM to 2:00 PM, the multimodal data stream from 1:00 AM to 2:00 PM is selected. Using the timestamp 2:00 AM as the mirror flip center point, the data stream is flipped to obtain a pseudo-data stream from 2:00 AM to 3:00 AM. The data at 2:00 AM in the multimodal data stream is equal to the data at 2:00 AM in the pseudo-data stream; the data at 1:30 AM in the multimodal data stream is equal to the data at 2:30 AM in the pseudo-data stream; the data at 1:00 AM in the multimodal data stream is equal to the data at 3:00 AM in the pseudo-data stream. This process is repeated to construct the initial pseudo-data stream for the vehicle's preset time period.

[0036] Step S203: Invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream. Use the pseudo data stream to initialize the Kalman filter, and use the initialized Kalman filter to perform fusion positioning processing on the pseudo data stream to obtain the driving trajectory corresponding to the preset time period. Then, perform time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle in the target time period.

[0037] The Kalman filter initialization (obtaining initial values) methods include the following two schemes:

[0038] 1) Use the fused positioning information (real-time forward filtering positioning on the vehicle) at the end of the original multimodal data stream when there is a positioning signal, that is, the Kalman filter is initialized at the beginning of the flipped pseudo data stream;

[0039] 2) Instead of using the vehicle's fused positioning results, use RTK and / or Lidar matched positioning information (because it is at the moment when there is a positioning signal, so there is RTK or Lidar matched positioning) to obtain the initial value of the Kalman filter.

[0040] In this embodiment, by applying steps S201, S202, and S203, after the vehicle regains its positioning signal, the historical multimodal data stream is mirrored and flipped with that moment as the time center. This generates a reverse pseudo-data stream that conforms to physical laws and is input into an unmodified forward Kalman filter, achieving "post-event vehicle trajectory reconstruction" without altering any underlying filtering architecture. This method completely avoids the highly invasive drawback of traditional bidirectional filters requiring dual-system maintenance. By simply reconstructing the spatiotemporal data layer, the original filter can convert the reversed data into a true forward input, achieving zero errors and instantaneous convergence. This not only significantly reduces system upgrade costs but also, for the first time, achieves high-precision reconstruction of the complete trajectory of "no-positioning-signal sections" (such as underground parking garages) in offline post-processing scenarios. It fills the trajectory gap in the initialization period of traditional positioning systems, providing data integrity support for autonomous driving algorithm training, ground truth generation, and system evaluation. This solves the problem that existing technologies require adding a reverse filter to achieve vehicle positioning in no-signal scenarios using forward and reverse filters, resulting in high positioning costs.

[0041] In the specific implementation process, the data values ​​in the initial pseudo data stream are inverted to obtain the pseudo data stream, including: identifying the first-order time-varying derivative in the initial pseudo data stream, wherein the first-order time-varying derivative includes at least one of the vehicle's RTK speed, chassis angular velocity, IMU angular velocity, and estimated gyroscope zero bias; and performing the inversion process on the first-order time-varying derivative in the initial pseudo data stream to obtain the pseudo data stream.

[0042] In addition, no changes are made to the data content for data that is approximately a single frame instantaneous data (data that is independent of time), including Lidar data, image data, RTK location data, etc.

[0043] In this embodiment, the first-order time-varying derivatives (such as RTK velocity, IMU angular velocity, and gyroscope bias) are physically inverted and reversed, while the second-order derivative (acceleration) remains unchanged, constructing an "asymmetric kinematic mapping" that conforms to the physical conservation laws of calculus. This design is based on the mechanical principle that "velocity direction reverses with time, but the force of acceleration remains essentially unchanged," ensuring that the reverse data stream does not trigger the anomaly detection mechanism in the Kalman filter. This avoids the situation where a complete flip would cause the IMU acceleration and CAN chassis acceleration directions to be abnormal, leading to abnormal filter rejection or degraded operation. This solution, by finely distinguishing the processing logic of the first and second derivatives, makes the pseudo-data stream physically "self-consistent and closed-loop," greatly improving the filter's acceptance of reconstructed data and its convergence stability. This embodiment achieves more robust and reliable trajectory reconstruction than pure time-domain flipping without any intrusion, enabling the system to maintain high-precision output even under complex operating conditions.

[0044] Specifically, the process of inverting the first-order time-varying derivative in the initial pseudo-data stream to obtain the pseudo-data stream includes: identifying the target gear information of the initial pseudo-data stream, wherein the target gear information includes at least one of the vehicle's parking gear, low gear, forward gear, and neutral gear; modifying the target gear information in the initial pseudo-data stream to reverse gear, and modifying the reverse gear in the initial pseudo-data stream to forward gear; and performing the inverting the first-order time-varying derivative in the initial pseudo-data stream to obtain the pseudo-data stream.

[0045] In this embodiment, vehicle gear information (DRIVE, PARK, etc.) is further mapped to REVERSE to resolve chassis control logic conflicts caused by data reversal. Kalman filters often dynamically adjust fusion weights or motion models based on gear information, such as assuming forward acceleration in "D" gear and enabling a reversing dynamics model in "R" gear. If the reverse data still retains the original gear, the filter will misjudge it as "reversing but using a forward model," leading to divergent state estimation. This solution uses gear logic deception to ensure that the filter still calls the matching dynamic prior in the reverse data stream, avoiding convergence failure caused by model mismatch. Although this processing is not universal, gear-assisted positioning is widely used in mainstream autonomous driving platforms and has strong engineering value. It significantly improves the system's compatibility and robustness in complex control logic environments, allowing pseudo-data streams to be "seamlessly integrated" into existing filters without additional modification to control logic, achieving zero-intrusion reconstruction that is indistinguishable from the real thing.

[0046] More specifically, after acquiring the multimodal data stream of the vehicle within the target time period, the method further includes: identifying the vehicle stationary periodic data stream in the multimodal data stream; and performing a cropping process on the vehicle stationary periodic data stream in the multimodal data stream to obtain a processed multimodal data stream.

[0047] In this embodiment, by identifying and cropping vehicle stationary periodic data, the amount of input data is significantly reduced, improving post-processing efficiency. Traditional methods require processing the entire acquired data segment, where a large number of stationary segments (such as waiting in a garage) have information entropy close to zero, yet still consume storage, computing, and transmission resources. This solution, through intelligent cropping, retains only the effective motion window, shortening the length of the input data stream and significantly reducing the computing power consumption and disk I / O pressure of the post-processing cluster. More importantly, eliminating stationary segments avoids their interference with the selection of the mirror center point, making the time reversal operation more accurate. Its technical effect is not only resource saving but also improved signal-to-noise ratio in the entire trajectory reconstruction process—after removing meaningless data, the filter can focus more on real motion characteristics, accelerating convergence and improving trajectory smoothness.

[0048] Furthermore, identifying the vehicle stationary periodic data stream in the aforementioned multimodal data stream includes: identifying the vehicle stationary periodic data stream in the aforementioned multimodal data stream by performing time-domain and frequency-domain detection on the IMU data in the aforementioned multimodal data stream; and / or, identifying the vehicle stationary periodic data stream in the aforementioned multimodal data stream by detecting the vehicle wheel speed data in the aforementioned multimodal data stream.

[0049] In this embodiment, two stationary segment identification methods are proposed: joint IMU time-domain and frequency-domain detection, or wheel speed threshold detection, constructing a highly robust stationary segment discrimination mechanism. IMU time-domain variance detection identifies low-amplitude fluctuations, while frequency-domain FFT analysis suppresses low-frequency gravity interference and high-frequency vibration interference, forming a dual filter to avoid misjudging the vehicle's stationary state. Wheel speed detection, as an auxiliary means, directly reflects the vehicle's motion state, forming redundant verification with the IMU. The two methods can be used independently or in combination, adapting to different sensor configuration scenarios. This method significantly improves the accuracy of stationary segment detection, avoiding the clipping of effective data. It provides precise anchor points for time compression, ensuring the correct "start and end point" logic for subsequent mirror flipping, and avoiding trajectory breakpoints or splicing distortion due to incorrect clipping. This identification mechanism becomes the key preprocessing scheme of this method, forming the foundation for achieving "high precision, high efficiency, and low error" trajectory reconstruction.

[0050] Furthermore, after obtaining the vehicle's driving trajectory within the target time period, the method further includes: evaluating and training the vehicle's localization algorithm using the driving trajectory, and evaluating and training the vehicle's perception algorithm.

[0051] In this embodiment, the reconstructed complete trajectory is used as a "virtual ground truth" for algorithm training and evaluation, solving the problem of "no data for scenarios without ground truth" in autonomous driving system development. Traditional methods rely on manual annotation or high-cost RTK ground truth vehicle data collection, which cannot cover weak signal areas such as tunnels and underground parking garages. This solution generates continuous trajectories in these areas through reverse reconstruction, which can be used to train perception models (such as LiDAR point cloud tracking), optimize localization fusion strategies, and verify the robustness of SLAM systems. This greatly expands the spatiotemporal coverage dimension of training data, enabling the algorithm to obtain real motion priors in the high-frequency fault scenario of "signal loss-recovery," improving the system's generalization ability in real urban environments. At the same time, this trajectory can be used as an evaluation metric to quantify key performance aspects such as positioning drift and initialization delay.

[0052] Specifically, taking the end timestamp of the aforementioned multimodal data stream as the center point, a mirror flip operation is performed on the aforementioned multimodal data stream to obtain the initial pseudo data stream for the aforementioned vehicle preset time period, including: using the formula: The above multimodal data stream is mirrored and flipped to obtain the above initial pseudo data stream for the target time period of the vehicle, wherein, The aforementioned end timestamp of the multimodal data stream, i.e., the aforementioned center point, The timestamps of the original data in the aforementioned multimodal data stream are as follows: This is the timestamp of the flipped data of the initial pseudo data stream mentioned above.

[0053] In this embodiment, a formula is used to achieve precise time mirroring, ensuring that all multimodal data streams are strictly symmetrically reconstructed on the time axis. This formula guarantees that the original time series and the reverse series are perfectly symmetrically distributed with the "end timestamp" as the mirror center, maintaining not only the consistency of event timing logic but also minimizing the clock synchronization error of each sensor. In heterogeneous data fusion scenarios such as GNSS, IMU, CAN, and LiDAR, time alignment is a prerequisite for fusion accuracy. The "centrally symmetrical mapping" achieved by this formula ensures that all data sources maintain their original time difference relationship after reversal, avoiding fusion drift caused by timestamp misalignment.

[0054] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the vehicle trajectory positioning method in the absence of positioning signals will be described in detail below with reference to specific embodiments.

[0055] The input sources for vehicle localization include lidar map matching and RTK (Real-Time Dynamic Differential Localization). Lidar matching algorithms have initialization capabilities and post-initialization tracking capabilities. When a vehicle starts in an underground parking garage with a lidar map, localization within the garage requires initialization. However, because lidar matching initialization relies on prior information from RTK, and there is no RTK signal within the garage, vehicle localization cannot be initialized until the vehicle leaves the garage and enters an area with an RTK signal. At this point, vehicle localization is successfully initialized through either RTK or lidar matching. Afterward, by continuously using Kalman filtering to fuse RTK, lidar matching, IMU wheel, and other observations, high-precision vehicle localization can be continuously obtained.

[0056] During the above process, the vehicle location can be obtained after the RTK signal is received after the vehicle leaves the garage, but the vehicle location during the time from starting the vehicle to leaving the garage cannot be obtained.

[0057] This embodiment relates to a specific vehicle trajectory localization method in a scenario without a positioning signal. It proposes a "data layer spatiotemporal mapping interception architecture" deployed at the front of the system, thereby achieving perfect and high-precision fast initial alignment without modifying any of the internal logic of the underlying mature forward filter (zero intrusion). At the same time, this embodiment also avoids the problem of existing solutions relying on extremely long periods of absolutely static data before start-up to gradually accumulate and converge to achieve localization, which leads to a huge waste of the simulation computing power and disk throughput of the system verification cluster. Figure 3 This is a flowchart illustrating a vehicle trajectory localization method in a specific scenario without a location signal, such as... Figure 3 As shown, it includes the following:

[0058] 1) Redundant Static Segment Removal and Time Warp Compression Module: Detects the required intervals at the beginning and end of the original message event stream. After establishing the basic vehicle attitude, it actively trims a large number of useless static cycles in the middle. The data at both ends are repositioned relative to each other using the center cut-off point as the boundary, and a new timestamp is assigned for seamless splicing.

[0059] Among them, the useless stationary period is determined by the data of IMU and / or wheel speed. Option 1: whether the wheel speed output by the vehicle is 0; Option 2: by detecting the time domain and frequency domain of IMU, and judging whether it is stationary according to a certain threshold.

[0060] 2) Time Reversal module: Extracts a data stream segment that effectively reflects the convergence state, and identifies the maximum boundary time of each sensor in the original time stream as the mapping center time point. ; through mapping formula The time series and array stream storage directions of the data are completely reversed. The timestamp of the data before the mirror flip. This is the timestamp of the data after mirroring and flipping; among them, the data stream time period that can well reflect the convergence state is extracted. This is output by the online positioning module. The positioning module will output an integrity monitoring index, which shows the current positioning status, such as uninitialized state, initialization completed state, and filter convergence completed state.

[0061] 3) Kinematic Multi-Decoupling Spatial Reversal Module: For time-reversed states, mathematically it's not just about reversing timestamps. This module finely processes physical domain variables: it performs pure physics-based inversion on the first-order dynamic derivatives of all dimensions involved in the system (such as RTK velocity, chassis angular velocity, IMU native angular velocity, and estimated gyroscope bias). and For the second-order time-varying derivatives of kinematics (such as the longitudinal / horizontal acceleration of the chassis measurement and the native linear acceleration of the IMU), according to the principle of calculus cancellation (the denominator time and the numerator velocity are reversed and remain unchanged), all acceleration characteristic constants are kept completely unreversed and unchanged.

[0062] 4) Vehicle control state / action relationship logic retrieval module: Based on the "backward deduction" fact constructed by the kinematic multi-decoupled spatial mapping module, this module remaps all the parking, low speed, drive, and neutral gear readings from the vehicle chassis CAN network and assigns them to motion signals such as "reverse gear".

[0063] 5) Fusion Input: This batch of modified and reset new hybrid data streams flows as "real-world" feedforward data into the unmodified classic Kalman filter. Since the data model has achieved complete physical consistency in the four pre-modules mentioned above, the system filter processes this data with zero errors, thus converging instantaneously and outputting the vehicle's driving trajectory in a no-signal scenario. The time axis of the obtained driving trajectory is a mirror-reversed time axis, which needs to be mapped according to the mirror-reversal relationship: The timeline corresponding to the driving trajectory is flipped and restored to obtain the vehicle driving trajectory with the correct timeline. This driving trajectory can be used for training localization algorithms, evaluating localization algorithms, training perception algorithms, and other offline applications that require vehicle trajectory tracking.

[0064] The embodiments of this application achieve the following technical effects:

[0065] 1) Achieving unprecedented "zero-intrusive code adaptation development": If the existing technical solution changes the positioning strategy or modifies and updates the dynamics transfer matrix and error equations, the entire backward EKF module needs to be rebuilt from scratch. According to this application, only this purely physical logic processing and time compression masking are required at the signal data input point. The forward Kalman filter architecture is completely unaware that the input data is undergoing inversion initialization processing, naturally ensuring that all forward error observation derivations are seamlessly closed-loop and effective.

[0066] 2) A highly robust virtual trajectory chain with kinematic error prevention was constructed: This application does not completely reverse or invert the numerical values, but innovatively utilizes the design theory that "acceleration, as the second derivative, should not be inverted." This reasoning ensures that even if forward driving is reversed, the reverse acceleration is still considered as a reasonable equivalent physical domain for forward braking (because the direction of the acceleration force characteristics does not change with time), avoiding the degradation and rejection caused by abnormal IMU parameters or CAN torque during the self-test of traditional positioning modules.

[0067] 3) Greatly releases the computing power of the algorithm: The Time Warp static segment merging method used skips non-essential information source processing, reducing the waiting time and computing load turnover of the entire offline processing pipeline by several times.

[0068] This application also provides a vehicle trajectory positioning device for scenarios without a positioning signal. It should be noted that the vehicle trajectory positioning device for scenarios without a positioning signal provided in this application can be used to execute the vehicle trajectory positioning method for scenarios without a positioning signal provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0069] The following describes the vehicle trajectory positioning device in the absence of positioning signals provided in the embodiments of this application.

[0070] Figure 4 This is a schematic diagram of a vehicle trajectory positioning device in a scenario without a positioning signal, according to an embodiment of this application. Figure 4 As shown, the device includes:

[0071] The acquisition unit 41 is used to acquire the multimodal data stream of the vehicle within a target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle drives out of the scenario without a positioning signal to when there is a positioning signal.

[0072] The mirror flipping operation unit 42 is used to perform a mirror flipping operation on the multimodal data stream with the end timestamp of the multimodal data stream as the center point to obtain the initial pseudo data stream of the vehicle preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream.

[0073] Processing unit 43 is used to invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream, use the pseudo data stream to initialize the Kalman filter, and use the initialized Kalman filter to perform fusion positioning processing on the pseudo data stream to obtain the driving trajectory corresponding to the preset time period. Then, according to the mirror mapping relationship of the mirror flip operation, the driving trajectory corresponding to the preset time period is restored by time axis processing to obtain the driving trajectory of the vehicle in the target time period.

[0074] In this embodiment, the acquisition unit is used to acquire the multimodal data stream of the vehicle within a target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle leaves the scenario without a positioning signal to when it enters the scenario with a positioning signal; the mirror flipping operation unit is used to perform a mirror flipping operation on the multimodal data stream with the end timestamp of the multimodal data stream as the center point to obtain the initial pseudo data stream of the vehicle within a preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream; the processing unit is used to invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream, and input the pseudo data stream into a Kalman filter to perform fusion positioning processing on the pseudo data stream using the Kalman filter to obtain the vehicle's driving trajectory within the target time period. By mirroring the historical multimodal data stream after the vehicle regains its positioning signal, using that moment as the time center, a physically-compliant reverse pseudo-data stream is generated and input into an unmodified forward Kalman filter. This achieves "post-event vehicle trajectory reconstruction" without altering any underlying filtering architecture. This method completely avoids the highly invasive nature of traditional bidirectional filters, which require dual-system maintenance. By reconstructing the spatiotemporal data layer, the original filter can convert the reversed data into a true forward input, achieving zero errors and instantaneous convergence. This not only significantly reduces system upgrade costs but also, for the first time, achieves high-precision reconstruction of the complete trajectory in "no-positioning-signal sections" (such as underground parking garages) in offline post-processing scenarios. This fills the trajectory gap in traditional positioning systems during the initialization period, providing data integrity support for autonomous driving algorithm training, ground truth generation, and system evaluation. This solves the problem of existing technologies requiring an additional reverse filter for vehicle positioning in no-signal scenarios, resulting in high positioning costs.

[0075] As an optional solution, the processing unit further includes a first identification module and a reversal processing module; the first identification module is used to identify the first-order time-varying derivative in the initial pseudo-data stream, wherein the first-order time-varying derivative includes at least one of the vehicle's RTK speed, chassis angular velocity, IMU angular velocity, and estimated gyroscope zero bias; the reversal processing module is used to perform the reversal processing on at least the first-order time-varying derivative in the initial pseudo-data stream to obtain the pseudo-data stream.

[0076] An optional scheme, the inversion processing module includes an identification submodule and a modification submodule; the identification submodule is used to identify the target gear information of the initial pseudo data stream, wherein the target gear information includes at least one of the vehicle's parking gear, low gear, forward gear, and neutral gear; the modification submodule is used to modify the target gear information of the initial pseudo data stream to reverse gear, and to modify the reverse gear in the initial pseudo data stream to forward gear, and to perform the inversion processing on the first-order time-varying derivative in the initial pseudo data stream to obtain the pseudo data stream.

[0077] In one optional embodiment, the apparatus further includes an identification unit and a trimming processing unit; the identification unit is used to identify the vehicle stationary periodic data stream in the multimodal data stream after acquiring the multimodal data stream within the target time period of the vehicle; the trimming processing unit is used to trim the vehicle stationary periodic data stream in the multimodal data stream to obtain a processed multimodal data stream.

[0078] In one optional scheme, the identification unit includes a second identification module and a third identification module; the second identification module is used to identify the vehicle stationary periodic data stream in the multimodal data stream by performing time-domain and frequency-domain detection on the IMU data in the multimodal data stream; the third identification module is used to identify the vehicle stationary periodic data stream in the multimodal data stream by detecting the vehicle wheel speed data in the multimodal data stream.

[0079] In one alternative embodiment, the apparatus further includes a processing unit, configured to, after obtaining the vehicle's driving trajectory within the target time period, evaluate and train the vehicle's positioning algorithm using the driving trajectory, and evaluate and train the vehicle's perception algorithm.

[0080] In one alternative, the mirror flipping operation unit includes a mirror flipping operation module, used to perform the following operation using the formula: The above multimodal data stream is mirrored and flipped to obtain the above initial pseudo data stream for the target time period of the vehicle, wherein, The aforementioned end timestamp of the multimodal data stream, i.e., the aforementioned center point, The timestamps of the original data in the aforementioned multimodal data stream are as follows: This is the timestamp of the flipped data of the initial pseudo data stream mentioned above.

[0081] The vehicle trajectory positioning device in the aforementioned scenario without a positioning signal includes a processor and a memory. The acquisition unit, mirror flipping operation unit, processing unit, etc., are all stored as program units in the memory, and the processor executes the program units stored in the memory to achieve the corresponding functions. All of the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0082] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured. By adjusting the kernel parameters, the problem of high cost associated with existing vehicle localization solutions in signal-free scenarios (which require adding a reverse filter) can be addressed.

[0083] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0084] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the vehicle trajectory positioning method in the scenario without a positioning signal.

[0085] Specifically, vehicle trajectory localization methods in scenarios without positioning signals include:

[0086] Step S201: Obtain the multimodal data stream of the vehicle within the target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle drives out of the scenario without a positioning signal to when there is a positioning signal.

[0087] Step S202: Using the end timestamp of the multimodal data stream as the center point, perform a mirror flip operation on the multimodal data stream to obtain the initial pseudo data stream of the vehicle's preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream.

[0088] Step S203: Invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream. Use the pseudo data stream to initialize the Kalman filter, and use the initialized Kalman filter to perform fusion positioning processing on the pseudo data stream to obtain the driving trajectory corresponding to the preset time period. Then, perform time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle in the target time period.

[0089] This invention provides a processor for running a program, wherein the program executes the vehicle trajectory positioning method in the scenario without positioning signal.

[0090] Specifically, vehicle trajectory localization methods in scenarios without positioning signals include:

[0091] Step S201: Obtain the multimodal data stream of the vehicle within the target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle drives out of the scenario without a positioning signal to when there is a positioning signal.

[0092] Step S202: Using the end timestamp of the multimodal data stream as the center point, perform a mirror flip operation on the multimodal data stream to obtain the initial pseudo data stream of the vehicle's preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream.

[0093] Step S203: Invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream. Use the pseudo data stream to initialize the Kalman filter, and use the initialized Kalman filter to perform fusion positioning processing on the pseudo data stream to obtain the driving trajectory corresponding to the preset time period. Then, perform time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle in the target time period.

[0094] This invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps:

[0095] Step S201: Obtain the multimodal data stream of the vehicle within the target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle drives out of the scenario without a positioning signal to when there is a positioning signal.

[0096] Step S202: Using the end timestamp of the multimodal data stream as the center point, perform a mirror flip operation on the multimodal data stream to obtain the initial pseudo data stream of the vehicle's preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream.

[0097] Step S203: Invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream. Use the pseudo data stream to initialize the Kalman filter, and use the initialized Kalman filter to perform fusion positioning processing on the pseudo data stream to obtain the driving trajectory corresponding to the preset time period. Then, perform time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle in the target time period.

[0098] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0099] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0100] Step S201: Obtain the multimodal data stream of the vehicle within the target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle drives out of the scenario without a positioning signal to when there is a positioning signal.

[0101] Step S202: Using the end timestamp of the multimodal data stream as the center point, perform a mirror flip operation on the multimodal data stream to obtain the initial pseudo data stream of the vehicle's preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream.

[0102] Step S203: Invert the data values ​​in the initial pseudo data stream to obtain a pseudo data stream. Use the pseudo data stream to initialize the Kalman filter, and use the initialized Kalman filter to perform fusion positioning processing on the pseudo data stream to obtain the driving trajectory corresponding to the preset time period. Then, perform time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle in the target time period.

[0103] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0104] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied 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.

[0105] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0106] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0109] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0110] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0111] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0112] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0113] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for vehicle trajectory localization in a scenario without a positioning signal, characterized in that, include: Acquire a multimodal data stream of the vehicle within a target time period, wherein the end timestamp of the multimodal data stream is the moment when the vehicle receives a positioning signal, and the target time period is the time period from when the vehicle leaves a scenario without a positioning signal to when it has a positioning signal. Using the end timestamp of the multimodal data stream as the center point, a mirror flip operation is performed on the multimodal data stream to obtain the initial pseudo data stream of the vehicle in a preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream. The data values ​​in the initial pseudo-data stream are inverted to obtain a new pseudo-data stream. This pseudo-data stream is then used to initialize a Kalman filter. The initialized Kalman filter is then used to perform fusion localization processing on the pseudo-data stream to obtain the driving trajectory corresponding to the preset time period. The driving trajectory corresponding to the preset time period is then restored according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle within the target time period. The driving trajectory is used to evaluate and train the vehicle's localization algorithm and its perception algorithm.

2. The method according to claim 1, characterized in that, The data values ​​in the initial pseudo-data stream are reversed to obtain the pseudo-data stream, which includes: Identify the first-order time-varying derivative in the initial pseudo data stream, wherein the first-order time-varying derivative includes at least one of the vehicle's RTK speed, chassis angular velocity, IMU angular velocity, and estimated gyroscope zero bias; The pseudo data stream is obtained by inverting at least the first-order time-varying derivative in the initial pseudo data stream.

3. The method according to claim 2, characterized in that, At least the first-order time-varying derivative in the initial pseudo-data stream is inverted and reversed to obtain the pseudo-data stream, including: Identify the target gear information of the initial pseudo data stream, wherein the target gear information includes at least one of the vehicle's parking gear, low gear, drive gear, and neutral gear; The target gear information in the initial pseudo data stream is modified to reverse gear, and the reverse gear in the initial pseudo data stream is modified to forward gear. The first-order time-varying derivative in the initial pseudo data stream is inverted to obtain the pseudo data stream.

4. The method according to claim 1, characterized in that, After acquiring the multimodal data stream of the vehicle within the target time period, the method further includes: Identify the vehicle stationary periodic data stream within the multimodal data stream; The vehicle stationary periodic data stream in the multimodal data stream is pruned to obtain the processed multimodal data stream.

5. The method according to claim 4, characterized in that, Identifying the vehicle stationary periodic data stream within the multimodal data stream includes: By performing time-domain and frequency-domain detection on the IMU data in the multimodal data stream, the vehicle stationary periodic data stream in the multimodal data stream can be identified; And / or, By detecting vehicle wheel speed data in the multimodal data stream, the vehicle stationary periodic data stream in the multimodal data stream is identified.

6. The method according to claim 1, characterized in that, Using the end timestamp of the multimodal data stream as the center point, a mirror flip operation is performed on the multimodal data stream to obtain the initial pseudo data stream of the vehicle within a preset time period, including: Through the formula: The multimodal data stream is mirrored and flipped to obtain the initial pseudo-data stream of the vehicle during the target time period, wherein... The center point, The original data timestamp of the multimodal data stream. The timestamp of the flipped data in the initial pseudo data stream.

7. A vehicle trajectory positioning device for scenarios without positioning signals, characterized in that, include: The acquisition unit is used to acquire the multimodal data stream of the vehicle within a target time period, wherein the end timestamp of the multimodal data stream is the time when the vehicle receives the positioning signal, and the target time period is the time period from when the vehicle leaves the scenario without a positioning signal to when it has a positioning signal. The mirror flipping operation unit is used to perform a mirror flipping operation on the multimodal data stream with the end timestamp of the multimodal data stream as the center point to obtain the initial pseudo data stream of the vehicle in a preset time period, wherein the beginning data of the initial pseudo data stream is the end data of the multimodal data stream, and the end data of the initial pseudo data stream is the beginning data of the multimodal data stream. The processing unit is used to invert the data values ​​in the initial pseudo-data stream to obtain a pseudo-data stream, initialize the Kalman filter using the pseudo-data stream, and perform fusion positioning processing on the pseudo-data stream through the initialized Kalman filter to obtain the driving trajectory corresponding to the preset time period. It then performs time axis recovery processing on the driving trajectory corresponding to the preset time period according to the mirror mapping relationship of the mirror flip operation to obtain the driving trajectory of the vehicle within the target time period. The driving trajectory is used to evaluate and train the vehicle's positioning algorithm and its perception algorithm.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the vehicle trajectory positioning method in a scenario without a positioning signal as described in any one of claims 1 to 6.

9. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing vehicle trajectory localization in a scenario without a localization signal as described in any one of claims 1 to 6.

Images