Vehicle cluster tunnel access positioning method based on intelligent cooperation and vehicle-mounted terminal

Abstract

The invention provides a vehicle cluster tunnel access positioning method based on intelligent cooperation and a vehicle-mounted terminal, relates to the technical field of intelligent driving control, and solves the problems of positioning interruption, inertial navigation error accumulation and insufficient cooperative positioning reliability caused by satellite signal failure of a vehicle cluster in a tunnel scene. The method comprises the following steps: judging whether each vehicle enters or exits from a tunnel area, when part of vehicles enter a tunnel, dynamically selecting a reference node from vehicles with effective satellite signals outside the tunnel, and broadcasting original observation values of the satellite signals; establishing a relative position constraint model of a vehicle cluster, acquiring relative position data of adjacent vehicles in real time among the vehicles, and performing dynamic weighted fusion on observation values, IMU data and relative position data of the vehicles with effective satellite signals by adopting a combined filtering algorithm; the satellite signal failure vehicle performs positioning calculation by combining the broadcast observation value with the IMU data and the relative position constraint model; and pre-capturing satellite signal loop parameters when driving out of the tunnel.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent driving control, and is applied to the process of vehicle cluster positioning. Specifically, it relates to a method for positioning the entry and exit of a vehicle cluster tunnel based on intelligent collaboration and an on-vehicle terminal. Background Art

[0002] With the rapid development of intelligent driving technology, vehicle clusters including freight fleets or passenger fleets are increasingly widely used in queue driving scenarios. In such scenarios, vehicles need to maintain stable formation and ensure driving safety through high-precision positioning technology. Especially in the open road environment with stable satellite signals, positioning technologies based on the Beidou Navigation System or the Global Navigation Satellite System (GNSS) can basically meet the requirements. However, in closed environments such as tunnels, satellite signals are completely invalidated due to physical obstruction, resulting in the problem of lost positioning reference for vehicle clusters relying solely on GNSS positioning. In the prior art, vehicles usually use an Inertial Navigation System (INS) for short-term positioning compensation, but the errors of inertial sensors will accumulate over time. Especially in long tunnel or complex curve scenarios, the positioning deviation relying solely on INS may exceed the lane width, leading to the risk of out-of-control safety distance between queue vehicles.

[0003] For tunnel scenarios on ordinary roads and highways, existing solutions mostly focus on single-vehicle positioning optimization. For example, pre-storing high-precision maps to match vehicle positions, or introducing lidar and vision sensors for feature recognition. However, the applicability of such methods in dynamically changing vehicle clusters is limited: on the one hand, differences in internal tunnel features such as lighting conditions and wall reflectivity may cause sensor misjudgment; on the other hand, single-vehicle positioning fails to fully utilize the collaborative information between vehicles, making it difficult to achieve positioning consistency for the entire cluster. In addition, although existing vehicle-road collaborative technologies attempt to provide auxiliary positioning through Road Side Units (RSUs), due to the multipath attenuation of communication signals in tunnels and deployment costs, their coverage and stability are difficult to meet the needs of large-scale vehicle clusters.

[0004] In scenarios lacking multi-source data fusion and collaborative positioning algorithms, the positioning robustness of vehicle clusters is significantly reduced. The errors of single-vehicle inertial navigation cannot be corrected by external information, and vehicle-vehicle collaboration based on simple communication protocols, such as periodic position broadcasting, also lacks dynamic evaluation of data credibility and is vulnerable to interference from abnormal nodes. At the same time, existing algorithms are mostly designed for static or low-speed scenarios and do not fully consider the dynamic characteristics such as sudden speed changes and formation adjustments when vehicle clusters enter and exit tunnels, resulting in the inability to match the positioning output with the actual motion state. These problems are particularly prominent in long-distance and high-load scenarios such as freight fleets, which may lead to queue disintegration or collision accidents.

[0005] In the prior art, there have also been studies attempting to combine UWB ultra-wideband or RFID beacons to enhance positioning within tunnels. However, this approach relies on pre-buried infrastructure, resulting in high implementation costs and poor cross-regional compatibility. Moreover, the Simultaneous Localization and Mapping (SLAM) technology based on vision or millimeter-wave radar is vulnerable to environmental interference within tunnels, with a high computational complexity, making it difficult to meet real-time requirements. Evidently, the prior art faces three major contradictions in the scenario of vehicle cluster tunnel entry and exit positioning: the contradiction between satellite signal failure and the need for positioning continuity, the contradiction between the perception limitations of individual vehicles and the need for cluster collaboration, and the contradiction between the assumption of a static environment and the dynamic traffic scenario. How to achieve low-cost and highly robust cluster collaborative positioning remains a technical bottleneck that urgently needs to be broken through in the field of intelligent driving. Summary of the Invention

[0006] The objective of the present invention is to solve the problems of positioning interruption caused by satellite signal failure, cumulative inertial navigation errors, and insufficient reliability of collaborative positioning for vehicle clusters in tunnel scenarios. Therefore, a method for vehicle cluster tunnel entry and exit positioning and an in-vehicle terminal based on intelligent collaboration are proposed. The present invention establishes a relative position constraint model for the vehicle cluster by fusing the original satellite signal observations, in-vehicle IMU inertial navigation data, and V2X position data, and uses an improved combined filtering algorithm to achieve dynamic weighted fusion of multi-source data. Meanwhile, when exiting the tunnel, the satellite signal is quickly reacquired through a pre-synchronization mechanism, significantly improving the positioning continuity inside and outside the tunnel, the collaborative accuracy of the cluster, and the adaptability to complex environments.

[0007] The present invention adopts the following technical solutions to achieve the objective:

[0008] A method for vehicle cluster tunnel entry and exit positioning based on intelligent collaboration includes the following steps:

[0009] S1. Obtain the position information of each vehicle in the vehicle cluster, and determine whether each vehicle enters or exits the tunnel area according to the sudden change amount of satellite signal strength and the pre-stored map data;

[0010] S2. When some vehicles in the vehicle cluster enter the tunnel, dynamically select at least one vehicle as a reference node from the vehicles with effective satellite signals outside the tunnel, and broadcast the original satellite signal observations to the vehicles with satellite signal failure inside the tunnel through V2X communication;

[0011] S3. During the process of the vehicle cluster entering the tunnel, establish a relative position constraint model for the vehicle cluster based on the queue topology relationship and kinematic equation of the vehicle cluster, and each vehicle obtains the relative position data of adjacent vehicles in real time through V2X communication;

[0012] S4. For vehicles with valid satellite signals, a combined filtering algorithm is used to dynamically weight and fuse the original satellite signal observations, in-vehicle IMU navigation data, and relative position data of adjacent vehicles to calculate their own positioning information until the satellite signals become invalid due to the vehicle entering a tunnel.

[0013] S5. For vehicles with invalid satellite signals, using the original satellite signal observations broadcast by the reference nodes as external observables, combined with their own in-vehicle IMU navigation data, based on the established relative position constraint model, calculate their own positioning information until the vehicle exits the tunnel.

[0014] S6. When some vehicles in the vehicle cluster exit the tunnel, the vehicles that have exited the tunnel broadcast the pre-generated ephemeris and frequency offset parameters to the vehicles still in the tunnel; the vehicles in the tunnel pre-configure the satellite signal acquisition loop parameters according to the ephemeris and frequency offset parameters and regain satellite positioning when they exit the tunnel.

[0015] Specifically, in step S1, to determine whether each vehicle enters or exits the tunnel area, it is as follows: when the drop amplitude of the carrier-to-noise ratio of the satellite signal received by the vehicle exceeds the first threshold within a preset time window, and the distance between its position information and the tunnel entrance coordinates in the pre-stored map data is less than the second threshold, it is determined that the vehicle enters the tunnel; when the distance between the inertial navigation estimated position of the vehicle and the tunnel exit coordinates in the pre-stored map data is less than the third threshold, and the carrier-to-noise ratio of the satellite signal recovers to above the fourth threshold, it is determined that the vehicle exits the tunnel.

[0016] Specifically, in step S2, the method of dynamically selecting reference node vehicles is as follows: determine the criteria for vehicles that can be used as reference node vehicles, including that the positioning dilution of precision of vehicles with valid satellite signals is less than a set threshold, the distance from vehicles in the tunnel does not exceed the maximum communication range, and they are in a preset proportion of the rear position in the vehicle cluster queue; the original satellite signal observations broadcast include Beidou / GNSS pseudorange, carrier phase value, and the corresponding satellite ephemeris data, and the broadcast action is executed according to a preset communication cycle.

[0017] Preferably, step S2 further includes a dynamic reference node switching mechanism: when the positioning dilution of precision of the original reference node vehicle exceeds the threshold or the distance from vehicles in the tunnel exceeds the maximum communication range, re-select a reference node from the remaining vehicles with valid satellite signals for switching; during the switching process, a weighted smooth transition method of historical broadcast data and new reference node data is adopted, and the transition duration is lower than a preset duration threshold.

[0018] Further, in step S3, the relative position constraint model is established as follows: based on the car-following model of the vehicle queue, the safety distance constraint and the heading angle consistency constraint between adjacent vehicles are defined. By using the relative distance and heading angle measurement values of adjacent vehicles obtained through V2X communication, combined with the preset sensor noise characteristics, a constraint equation is generated, and the constraint weight is dynamically updated based on the historical data within the sliding time window; among them, the weight of the heading angle consistency constraint is inversely proportional to the vehicle's steering angular velocity.

[0019] Preferably, when establishing the relative position constraint model, a lightweight compensation network is embedded in the model; the lightweight compensation network takes the historical positioning residual, the queue topology change rate, and the statistical characteristics of IMU errors as inputs, and outputs the position drift compensation amount of IMU navigation; the lightweight compensation network is obtained through offline training, and the training data includes sensor data and true value data for multi-vehicle cooperative positioning in a simulated tunnel scenario; the structure of the lightweight compensation network adopts three fully connected layers, and its parameter quantity is less than the preset parameter quantity threshold corresponding to real-time inference of in-vehicle terminals.

[0020] Further, in step S4, the combined filtering algorithm adopts an improved federated Kalman filter architecture, and local filters are respectively set for the original satellite signal observations, the in-vehicle IMU navigation data, and the relative position data of adjacent vehicles. Among them, the weight of the original satellite signal observations is dynamically adjusted according to the number of satellites and the geometric distribution factor, and the weight of the relative position data of adjacent vehicles is adaptively updated based on the residual covariance matrix of the relative position constraint model. Finally, the global state optimal estimation is achieved through the main filter.

[0021] Further, in step S5, the positioning calculation process of the vehicle with satellite signal failure is specifically as follows: converting the original satellite signal observations broadcast by the reference node vehicle into an equivalent pseudorange observation equation, and jointly constructing the state equation of the extended Kalman filter with the position calculated from its own in-vehicle IMU navigation data and the relative position constraint model; among them, the system noise covariance of IMU navigation increases exponentially according to the driving duration in the tunnel, while the observation noise covariance of the relative position constraint model decays dynamically according to the update frequency of adjacent vehicle data.

[0022] Specifically, in step S6, the generation method of the pre-generated ephemeris and frequency offset parameters is as follows: for the vehicle with effective satellite signals after driving out of the tunnel, the current satellite ephemeris is calculated by using the restored Beidou / GNSS signals, and combined with the time difference results of its own in-vehicle IMU navigation data and satellite observations, the Doppler frequency shift parameters of the satellite signals are predicted to generate a frequency offset compensation table; among them, the Doppler frequency shift parameters include the carrier frequency deviation and the code phase pre-compensation amount.

[0023] The present invention also provides an in-vehicle terminal for vehicle cluster tunnel entry and exit positioning, including:

[0024] A tunnel status detection module, which is used to obtain the position information of each vehicle in the vehicle cluster in real time, and determine whether each vehicle enters or exits the tunnel area according to the sudden change amount of satellite signal strength and the pre-stored map data;

[0025] A dynamic reference node management module, which is connected to the V2X communication module, and is used to select reference nodes from vehicles with effective satellite signals outside the tunnel when some vehicles in the vehicle cluster enter the tunnel, and control them to broadcast the original satellite signal observations;

[0026] A cooperative positioning processing module, including a relative position constraint modeling unit and a data fusion unit; the relative position constraint modeling unit is used to establish a relative position constraint model of the vehicle cluster during the process of the vehicle cluster entering the tunnel according to the vehicle queue topology relationship and kinematic equations; the data fusion unit uses a combined filtering algorithm to dynamically weight and fuse the original satellite signal observations, in-vehicle IMU navigation data and relative position data of adjacent vehicles, and solve the positioning information of vehicles with effective satellite signals outside the tunnel;

[0027] A signal failure positioning module, which is used to, when the satellite signal fails, use the original satellite signal observations broadcast by the reference node vehicle as an external input, and combine its own in-vehicle IMU navigation data and relative position constraint model to solve the positioning information of vehicles with satellite signal failures inside the tunnel;

[0028] A satellite signal pre-synchronization module, which is used to pre-configure the satellite signal acquisition loop according to the received ephemeris and frequency offset parameters when some vehicles in the vehicle cluster exit the tunnel, and trigger the rapid reacquisition of satellite signals;

[0029] A V2X communication module, which is used to interact with other in-vehicle terminals in the vehicle cluster in real time for the original satellite signal observations, relative position data of adjacent vehicles, ephemeris and frequency offset parameters according to a preset communication cycle.

[0030] In summary, due to the adoption of this technical solution, the beneficial effects of the present invention are as follows:

[0031] Through the intelligent selection of dynamic reference nodes and the multi-source data collaborative fusion mechanism, the present invention effectively solves the problem of positioning continuity of the vehicle cluster in the tunnel scenario. When the satellite signals of some vehicles in the vehicle cluster are partially invalid, the vehicles that have not entered the tunnel are used as dynamic reference nodes, and combined with their original satellite observations and inertial navigation data, the error accumulation speed of traditional inertial navigation is significantly suppressed, especially suitable for long tunnels or sections with dense curves. Compared with the scheme that relies on fixed roadside units or buried beacons, the present invention reduces the infrastructure dependence and implementation cost through the self-organizing cooperation of the vehicle cluster, and at the same time avoids the global positioning risk caused by the failure of a single node.

[0032] By introducing vehicle queue topology constraints and an improved combined filtering algorithm, the present invention enhances the overall accuracy and robustness of cluster positioning. The relative position constraint model converts the kinematic relationship between adjacent vehicles into mathematical constraint conditions, effectively correcting the drift error of inertial navigation. Especially in the middle section of the tunnel where satellite signals completely fail, centimeter-to-decimeter-level positioning accuracy can still be maintained. In addition, the dynamic weighted fusion strategy can adjust the data weights in real time according to satellite signal quality, communication stability, and sensor reliability, solving the problem of insufficient adaptability of traditional fixed-weight algorithms when switching between inside and outside the tunnel.

[0033] In the tunnel exit stage, the satellite signal fast reacquisition technology based on the pre-synchronization mechanism significantly shortens the positioning recovery time. By sharing ephemeris and frequency offset parameters between vehicles, the receiving vehicle can pre-configure the signal acquisition loop parameters, greatly shortening the satellite signal locking time and avoiding the formation disorder caused by positioning delay after the vehicle exits the tunnel. This method is not only applicable to straight queue scenarios but also can be extended to complex formation requirements with multiple lanes and variable formations, providing high-reliability positioning guarantee for the tunnel passage of intelligent driving vehicle clusters. Brief Description of the Drawings

[0034] Figure 1 It is a schematic diagram briefly showing the overall process of the vehicle cluster tunnel entry and exit positioning method of the present invention;

[0035] Figure 2 It is a schematic diagram showing the composition of the functional modules of the vehicle-mounted terminal of the present invention. Detailed Embodiments

[0036] 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. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0037] Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but merely represents selected 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 fall within the scope of protection of the present invention.

[0038] Embodiment 1

[0039] A vehicle cluster tunnel entry and exit positioning method based on intelligent collaboration, Figure 1 shows a brief overview of the overall process of this method. For synchronous reference, the key steps of this method are summarized as follows:

[0040] S1. Obtain the position information of each vehicle in the vehicle cluster, and determine whether each vehicle enters or exits the tunnel area according to the sudden change amount of satellite signal strength and the pre-stored map data;

[0041] S2. When some vehicles in the vehicle cluster enter the tunnel, dynamically select at least one vehicle as a reference node from the vehicles with effective satellite signals outside the tunnel, and broadcast the original satellite signal observations to the vehicles with invalid satellite signals in the tunnel through V2X communication;

[0042] S3. During the process of the vehicle cluster entering the tunnel, based on the queue topology relationship and kinematic equation of the vehicle cluster, establish a relative position constraint model of the vehicle cluster, and each vehicle obtains the relative position data of adjacent vehicles in real time through V2X communication;

[0043] S4. For vehicles with effective satellite signals, use a combined filtering algorithm to dynamically weight and fuse their original satellite signal observations, in-vehicle IMU navigation data, and relative position data of adjacent vehicles, and calculate their own positioning information until their own satellite signals become invalid due to entering the tunnel;

[0044] S5. For vehicles with invalid satellite signals, use the original satellite signal observations broadcast by the reference node as external observables, combine their own in-vehicle IMU navigation data, and calculate their own positioning information based on the established relative position constraint model until they exit the tunnel;

[0045] S6. When some vehicles in the vehicle cluster exit the tunnel, the vehicles that have exited the tunnel broadcast the pre-generated ephemeris and frequency offset parameters to the vehicles still in the tunnel; the vehicles in the tunnel pre-configure the satellite signal acquisition loop parameters according to the ephemeris and frequency offset parameters, and regain satellite positioning when they exit the tunnel.

[0046] When the above method is actually applied, when all vehicles in the vehicle cluster enter the tunnel and all satellite signals become invalid, this embodiment will use the original satellite observations broadcast by the last vehicle that lost satellite signals before the signal disappeared as a reference, and combine the queue kinematic relationships defined in the relative position constraint model, such as following safety distance, heading angle synchronization, etc. Each vehicle in the tunnel continuously exchanges relative position measurement values through V2X, and uses the historical constraint equation residuals within the sliding time window to dynamically correct the inertial navigation error.

[0047] At this stage, the combined filtering algorithm correlates the system noise covariance of inertial navigation with the observation noise covariance of the constraint model, adaptively adjusts according to the vehicle driving duration and the queue deformation rate, reducing the growth rate of the positioning error by more than 50% compared with traditional pure inertial navigation until some vehicles drive out of the tunnel and the satellite signal reference is restored. Moreover, when existing vehicles drive out of the tunnel and obtain satellite signals, they can act as reference nodes again as when entering the tunnel. Based on the pre-generated ephemeris and frequency offset parameters, they broadcast the original satellite signal observations to the vehicles in the tunnel at the same time, ensuring the positioning accuracy of the vehicles in the tunnel.

[0048] Next, this embodiment exemplarily illustrates the details of each step of the method.

[0049] In step S1, before each vehicle enters the tunnel, its on-vehicle terminal can continuously monitor the change in the carrier-to-noise ratio of the Beidou / GNSS signal. For example, when the signal strength drops suddenly from 45 dB-Hz to below 28 dB-Hz within 0.5 seconds and the deviation between the vehicle position and the tunnel entrance in the pre-stored high-precision map is less than 5 meters, it is determined that the vehicle has entered the tunnel area. The pre-stored map can adopt the vector layer format, including the three-dimensional contour of the tunnel, the curvature of the lane lines, and the slope information, and ensure the accuracy reaches the centimeter level. For the judgment of driving out of the tunnel, it can be comprehensively triggered by combining the distance between the position deduced by the inertial navigation IMU and the tunnel exit coordinates, for example, less than 10 meters, and the recovery of the satellite signal, for example, the carrier-to-noise ratio rises above 35 dB-Hz.

[0050] In step S2, after some vehicles enter the tunnel, the dynamic reference node selection algorithm is immediately started. The reference node needs to meet the positioning dilution of precision PDOP less than 2.5, the distance from the vehicles in the tunnel does not exceed the effective range of V2X communication, usually 300 meters, and the vehicle at the 30% position at the rear of the queue is preferably selected. The selected node broadcasts information such as the pseudorange observations of the Beidou / GNSS signal frequency points and the solution results of the L1 carrier phase integer ambiguity through its V2X communication module at a period of 20 ms. At the same time, it can also broadcast its own on-vehicle IMU navigation data. This data is pre-processed by Kalman filtering, and the three-dimensional position, speed, and attitude angle it includes can also facilitate the information communication and interaction between vehicles. The encapsulation of the broadcast data information follows the SAE J2735 standard, adding a timestamp and a data integrity check code.

[0051] During the process of the vehicle cluster in step S3 entering the tunnel, the relative position constraint model is constructed by the cooperative positioning processing module in the on-vehicle terminal. Taking a four-vehicle formation cluster as an example in this embodiment, based on the improved intelligent driver model IDM, the safe following distance d of adjacent vehicles is defined safe , and the calculation formula is as follows:

[0052]

[0053] where v n is the speed of the nth vehicle, v n-1 is the speed of the (n - 1)th vehicle, t hw is the preset headway time, which can be defaulted to 1.2 seconds, and b max is the maximum braking deceleration of the vehicle. At the same time, the relative distance and heading angle deviation of the preceding vehicle are obtained through millimeter-wave radar and V2X communication, and the heading consistency constraint is constructed by combining the angular velocity meter data of the IMU system. The noise covariance matrix of each constraint equation is dynamically set according to the sensor type; in this embodiment, the radar distance noise equation can be 0.01m 2 , the V2X heading angle noise variance can be 0.1rad 2 , and the IMU angular velocity noise variance is adaptively adjusted according to the temperature drift coefficient.

[0054] Preferably, in this embodiment, a lightweight compensation network is embedded in the constructed relative position constraint model; this network takes historical positioning residuals, queue topology change rates, and IMU error statistical characteristics as inputs, and outputs the position drift compensation amount of IMU navigation; this network is obtained through offline training, and the training data includes sensor data and true value data for multi-vehicle cooperative positioning in a simulated tunnel scenario; specifically, the structure can adopt three fully connected layers, and its parameter quantity is less than the preset parameter quantity threshold corresponding to supporting real-time inference of in-vehicle terminals.

[0055] In step S4, for vehicles with valid satellite signals, the multi-source data fusion process is started. The in-vehicle terminal first performs quality screening on the original Beidou / GNSS observations, eliminates satellite data with an elevation angle lower than 15 degrees, and uses the dual-frequency ionospheric delay correction technology to improve the pseudorange measurement accuracy. The inertial navigation data is sampled at a frequency of 200Hz, and the thermal drift effect of the gyro zero bias is eliminated through the temperature compensation module. At the same time, it is loosely combined with the wheel speed pulse information to preliminarily calculate the short-term motion trajectory of the vehicle. The relative position data broadcast by adjacent vehicles is input to the local filter of the federated filtering architecture after timestamp alignment and outlier filtering. The weight of the satellite observations is dynamically adjusted according to the current number of visible satellites and their spatial distribution. For example, when the number of satellites is greater than 8 and the azimuth angles are evenly distributed, the weight of the satellite data is increased to 70%, while the weight of the adjacent vehicle data is correspondingly reduced to 20%, and the remaining 10% is allocated to the inertial navigation calculation result, so as to balance the confidence differences of different sensors. The vehicle then calculates its own positioning information when there is a satellite signal.

[0056] In step S5, when the vehicle enters the tunnel and causes the satellite signal to completely fail, the positioning system switches to a tightly coupled mode with reference node data as the core. The pseudorange observations broadcast by the reference nodes are converted into geometric distance constraints with the position of the receiving vehicle. Combining the speed and acceleration information deduced by its own inertial navigation, as well as the established relative position constraint model, positioning estimation is carried out. During this process, the system error covariance matrix of inertial navigation exponentially amplifies with the driving duration in the tunnel. For example, the error covariance increases by 1.5 times every 10 seconds, while the observation noise of the relative position constraint model gradually decays according to the update frequency of neighboring vehicle data. If the neighboring vehicle data is stably updated every 50 ms, the observation noise variance can be reduced to 30% of the initial value. This dynamic adjustment mechanism enables the lateral positioning error to be controlled within a growth rate of 0.5 meters per minute in the long tunnel scenario, significantly superior to the 2-meter-per-minute error accumulation of traditional pure inertial navigation.

[0057] In step S6, during the stage when the vehicle exits the tunnel, the satellite signal pre-synchronization module takes effect. The first batch of vehicles that recover satellite signals complete the ephemeris parsing of the currently visible satellites within 0.5 seconds, and based on the time difference between inertial navigation data and the satellite signal carrier phase, estimate the Doppler frequency shift of each satellite. For example, for BDS-3GEO satellites, the pre-estimated value of the carrier frequency offset is accurate to ±3 Hz, and the error of the pre-compensation amount of the code phase is less than 0.2 chips. These parameters are broadcast to other vehicles in the queue through V2X. The receiving vehicle pre-configures the acquisition loop of the satellite receiver accordingly: compresses the carrier frequency search range from the conventional ±5 kHz to ±150 Hz, shortens the integration time from 20 ms to 5 ms, and loads the pre-compensated code phase offset. Experiments show that this mechanism can shorten the satellite signal re-acquisition time from an average of 2.8 seconds to within 0.9 seconds, ensuring the recovery of centimeter-level positioning accuracy within 10 seconds after the vehicle exits the tunnel and avoiding the formation from being disordered due to positioning delay.

[0058] Finally, for the scenario where the entire vehicle cluster queue is inside the tunnel in this embodiment, the pure relative positioning fault-tolerant mode can be activated. Based on the historical data of the last vehicle that lost satellite signals, the angular velocity integration error of inertial navigation is dynamically corrected using the heading angle consistency constraint residuals within a 5-second sliding time window. For example, if the heading angle measurement residuals exceed 0.3 degrees continuously for 3 times, the system automatically triggers the online calibration of the gyroscope zero bias, and at the same time increases the relative distance constraint weight by 40% to suppress position drift. In this mode, the lateral position error of the vehicle can be stabilized within 0.8 meters until at least one vehicle exits the tunnel and provides a new global positioning reference, realizing a smooth transition and recovery of the positioning state of the entire queue.

[0059] Embodiment 2

[0060] Based on Embodiment 1, this embodiment provides an in-vehicle terminal for vehicle cluster tunnel entry and exit positioning, and the composition of its functional modules can be referred toFigure 2 The schematic diagram of this on-vehicle terminal can serve as the hardware basis for the vehicle cluster tunnel entry and exit positioning method in Embodiment 1, thereby implementing the content of each step of this method. In this embodiment, the on-vehicle terminal is composed as follows:

[0061] A tunnel state detection module, which is used to obtain the position information of each vehicle in the vehicle cluster in real time, and judge whether each vehicle enters or exits the tunnel area according to the sudden change amount of satellite signal strength and pre-stored map data;

[0062] A dynamic reference node management module, which is connected to the V2X communication module, and is used to select reference nodes from the vehicles with effective satellite signals outside the tunnel when some vehicles in the vehicle cluster enter the tunnel, and control them to broadcast the original satellite signal observations;

[0063] A cooperative positioning processing module, including a relative position constraint modeling unit and a data fusion unit; the relative position constraint modeling unit is used to establish a relative position constraint model of the vehicle cluster during the process of the vehicle cluster entering the tunnel according to the vehicle queue topology relationship and kinematic equations; the data fusion unit uses a combined filtering algorithm to dynamically weight and fuse the original satellite signal observations, on-vehicle IMU navigation data, and relative position data of adjacent vehicles, and calculates the positioning information of the vehicles with effective satellite signals outside the tunnel;

[0064] A signal failure positioning module, which is used to calculate the positioning information of the vehicles with satellite signal failures in the tunnel by using the original satellite signal observations broadcast by the reference node vehicles as external inputs, combined with its own on-vehicle IMU navigation data and relative position constraint model when the satellite signals fail;

[0065] A satellite signal pre-synchronization module, which is used to pre-configure the satellite signal acquisition loop according to the received ephemeris and frequency offset parameters when some vehicles in the vehicle cluster exit the tunnel, and trigger the rapid reacquisition of satellite signals;

[0066] A V2X communication module, which is used to interact with other on-vehicle terminals in the vehicle cluster in real time for the original satellite signal observations, relative position data of adjacent vehicles, ephemeris, and frequency offset parameters according to a preset communication cycle.

Claims

1. A vehicle cluster tunnel entry and exit positioning method based on intelligent collaboration, characterized in that It includes the following steps: S1. Obtain the position information of each vehicle in the vehicle cluster, and judge whether each vehicle enters or exits the tunnel area according to the sudden change amount of satellite signal strength and the pre-stored map data; S2. When some vehicles in the vehicle cluster enter the tunnel, dynamically select at least one vehicle as a reference node from the vehicles with effective satellite signals outside the tunnel, and broadcast the original satellite signal observations to the vehicles with invalid satellite signals in the tunnel through V2X communication; S3. During the process of the vehicle cluster entering the tunnel, based on the queue topology relationship and kinematic equation of the vehicle cluster, establish a relative position constraint model of the vehicle cluster, and each vehicle obtains the relative position data of adjacent vehicles in real time through V2X communication; S4. For vehicles with effective satellite signals, use a combined filtering algorithm to dynamically weight and fuse the original satellite signal observations, on-vehicle IMU navigation data, and relative position data of adjacent vehicles, and calculate its own positioning information until its own satellite signal fails due to entering the tunnel; S5. For vehicles with invalid satellite signals, use the original satellite signal observations broadcast by the reference node as external observables, combine its own on-vehicle IMU navigation data, and calculate its own positioning information based on the established relative position constraint model until it exits the tunnel; S6. When some vehicles in the vehicle cluster exit the tunnel, the vehicles that have exited the tunnel broadcast the pre-generated ephemeris and frequency offset parameters to the vehicles still in the tunnel; the vehicles in the tunnel pre-configure the satellite signal acquisition loop parameters according to the ephemeris and frequency offset parameters, and regain satellite positioning when exiting the tunnel.

2. The vehicle cluster tunnel entry and exit positioning method according to claim 1, characterized in that In step S1, judging whether each vehicle enters or exits the tunnel area specifically is: when the drop amplitude of the carrier-to-noise ratio of the satellite signal received by the vehicle exceeds the first threshold within a preset time window, and the distance between its position information and the tunnel entrance coordinates in the pre-stored map data is less than the second threshold, it is determined that the vehicle enters the tunnel; when the distance between the inertial navigation estimated position of the vehicle and the tunnel exit coordinates in the pre-stored map data is less than the third threshold, and the carrier-to-noise ratio of the satellite signal recovers to above the fourth threshold, it is determined that the vehicle exits the tunnel.

3. The vehicle cluster tunnel entry and exit positioning method according to claim 1, characterized in that In step S2, the method of dynamically selecting the reference node vehicle is: determine the criteria for the vehicle that can be used as the reference node vehicle, including that the positioning dilution of precision of the vehicle with effective satellite signals is less than the set threshold, the distance from the vehicle in the tunnel does not exceed the maximum communication range, and it is in the position of a preset proportion at the rear in the vehicle cluster queue; the original satellite signal observations broadcast include Beidou / GNSS pseudorange, carrier phase value, and the corresponding satellite ephemeris data, and the broadcast action is executed according to the preset communication cycle.

4. The vehicle cluster tunnel entry and exit positioning method according to claim 3, characterized in that: Step S2 also includes a dynamic reference node switching mechanism: when the positioning dilution of precision of the original reference node vehicle exceeds the threshold or the distance from the vehicle in the tunnel exceeds the maximum communication range, re-select a reference node from the remaining vehicles with effective satellite signals for switching; during the switching process, a weighted smooth transition method of historical broadcast data and new reference node data is adopted, and the transition duration is lower than the preset duration threshold.

5. The vehicle cluster tunnel entry and exit positioning method according to claim 1, characterized in that In step S3, the relative position constraint model is established as follows: Based on the car-following model of the vehicle queue, the safety distance constraint and the heading angle consistency constraint between adjacent vehicles are defined. Using the relative distance and heading angle measurement values of adjacent vehicles obtained by V2X communication, combined with the preset sensor noise characteristics, constraint equations are generated, and the constraint weights are dynamically updated based on the historical data within a sliding time window; among them, the weight of the heading angle consistency constraint is inversely proportional to the vehicle's steering angular velocity.

6. The vehicle cluster tunnel entry and exit positioning method according to claim 5, characterized in that: When establishing the relative position constraint model, a lightweight compensation network is embedded in the model; the lightweight compensation network takes the historical positioning residual, the queue topology change rate, and the IMU error statistical characteristics as inputs, and outputs the position drift compensation amount of the IMU navigation; the lightweight compensation network is obtained through offline training, and the training data includes the sensor data and the true value data of multi-vehicle cooperative positioning in a simulated tunnel scenario; the structure of the lightweight compensation network adopts three fully connected layers, and its number of parameters is less than the preset parameter threshold corresponding to the real-time inference of the vehicle-mounted terminal.

7. The vehicle cluster tunnel entry and exit positioning method according to claim 1, characterized in that: In step S4, the combined filtering algorithm adopts an improved federated Kalman filter architecture, and local filters are set for the original satellite signal observations, the vehicle-mounted IMU navigation data, and the relative position data of adjacent vehicles respectively. Among them, the weight of the original satellite signal observations is dynamically adjusted according to the number of satellites and the geometric distribution factor, and the weight of the relative position data of adjacent vehicles is adaptively updated based on the residual covariance matrix of the relative position constraint model. Finally, the global state optimal estimation is achieved through the main filter.

8. The vehicle cluster tunnel entry and exit positioning method according to claim 1, characterized in that, In step S5, the positioning solution process of the vehicle with satellite signal failure is specifically as follows: The original satellite signal observations broadcast by the reference node vehicle are converted into equivalent pseudo-range observation equations, and jointly constructed with the position calculated from its own vehicle-mounted IMU navigation data and the relative position constraint model as the state equation of the extended Kalman filter; among them, the system noise covariance of the IMU navigation increases exponentially according to the driving duration in the tunnel, while the observation noise covariance of the relative position constraint model decays dynamically according to the update frequency of the adjacent vehicle data.

9. The vehicle cluster tunnel entry and exit positioning method according to claim 1, characterized in that, In step S6, the generation method of the pre-generated ephemeris and frequency offset parameters is as follows: For the vehicle with effective satellite signals after driving out of the tunnel, the current satellite ephemeris is calculated using the restored Beidou / GNSS signals, and combined with the time difference result of its own vehicle-mounted IMU navigation data and satellite observations, the Doppler frequency shift parameter of the satellite signal is predicted to generate a frequency offset compensation table; among them, the Doppler frequency shift parameter includes the carrier frequency deviation and the code phase pre-compensation amount.

10. A vehicle-mounted terminal for vehicle cluster tunnel entry and exit positioning, characterized in that, including: A tunnel state detection module, which is used to obtain the position information of each vehicle in the vehicle cluster in real time, and judge whether each vehicle enters or exits the tunnel area according to the sudden change amount of the satellite signal strength and the pre-stored map data; A dynamic reference node management module, which is connected to the V2X communication module, and is used to select a reference node from the vehicles with effective satellite signals outside the tunnel when some vehicles in the vehicle cluster enter the tunnel, and control it to broadcast the original satellite signal observations; A cooperative positioning processing module, including a relative position constraint modeling unit and a data fusion unit; The relative position constraint modeling unit is used to establish a relative position constraint model of the vehicle cluster during the process of the vehicle cluster entering the tunnel according to the vehicle queue topological relationship and kinematic equations; the data fusion unit uses a combined filtering algorithm to dynamically weight and fuse the original satellite signal observations, in-vehicle IMU navigation data, and relative position data of adjacent vehicles, and calculates the positioning information of vehicles with effective satellite signals outside the tunnel; The signal failure positioning module is used to calculate the positioning information of vehicles with satellite signal failure in the tunnel when the satellite signal fails, with the original satellite signal observations broadcast by the reference node vehicle as the external input, combined with its own in-vehicle IMU navigation data and relative position constraint model; The satellite signal pre-synchronization module is used to pre-configure the satellite signal acquisition loop according to the received ephemeris and frequency offset parameters and trigger the rapid reacquisition of satellite signals when some vehicles in the vehicle cluster drive out of the tunnel; The V2X communication module is used to interact with other in-vehicle terminals in the vehicle cluster in real time for the original satellite signal observations, relative position data of adjacent vehicles, ephemeris, and frequency offset parameters according to a preset communication cycle.

Images