High-precision beidou positioning terminal device and system

Abstract

The application relates to the field of satellite positioning technology and discloses a high-precision Beidou positioning terminal device and system, which comprises a Beidou multi-frequency positioning module, a heterogeneous data fusion access engine, a neuromorphic space-time alignment algorithm module, a blockchain trusted data pool module and a scene semantic decision module. The Beidou multi-frequency positioning module generates centimeter-level original observation data by receiving Beidou satellite B1C, B2a and B3I frequency band signals. The heterogeneous data fusion access engine aligns the space-time of multi-source sensing data by dynamically accessing laser point cloud, visual SLAM and UWB positioning data. The neuromorphic space-time alignment algorithm module unifies the coordinate systems of the multi-source data. The blockchain trusted data pool module anchors the Beidou positioning data and encrypts and stores the sensing data. The scene semantic decision module encodes the dynamic characteristics of a carrier phase into a pulse time sequence in real time through a neuromorphic module, converts a physical signal into a space-time event that can be understood by an algorithm, and automatically improves the weight of laser point cloud and simultaneously suppresses visual interference and reduces the positioning error in a satellite denial scenario through a cross-modal attention mechanism.

Description

TECHNICAL FIELD

[0001] The application relates to the field of satellite positioning technology, in particular to a high-precision Beidou positioning terminal device and system. BACKGROUND

[0002] With the deep application of location services in automatic driving, intelligent transportation and other fields, the industry puts forward three rigid requirements for the positioning system. In a complex electromagnetic environment (such as a densely built urban area and a strong industrial interference zone), centimeter-level positioning continuity is maintained, time and space reference of multi-source heterogeneous sensing data such as laser radar, visual sensor and wireless ranging is unified, and positioning evidence chain that can be judicially verified is provided for high-value scenarios (accident accountability, operation audit). However, the current solution has not systematically covered the full-link capabilities of signal anti-interference, multi-source fusion and trusted storage, which restricts the large-scale landing of high-precision positioning in safety-related scenarios.

[0003] However, in the current technology, the carrier phase dynamic characteristics are only used as numerical input and are not event-encoded, and the weight adaptive adjustment of multi-source sensing data cannot be performed in real time, resulting in errors in satellite positioning in different scenarios. SUMMARY

[0004] In view of the deficiencies of the prior art, the application provides a high-precision Beidou positioning terminal device and system, which solves the problem that it is difficult to encode in real time as a pulse time sequence, and the adaptive adjustment of the weight of multi-source sensing data cannot be triggered in real time, resulting in errors in satellite positioning in different scenarios.

[0005] To achieve the above purpose, the application is implemented by the following technical scheme: a high-precision Beidou positioning system, comprising:

[0006] A Beidou multi-frequency positioning module generates centimeter-level raw observation data by receiving Beidou satellite B1C, B2a and B3I frequency band signals;

[0007] An heterogeneous data fusion access engine time and space aligns multi-source sensing data by dynamically accessing laser point cloud, visual SLAM and UWB positioning data;

[0008] A neuromorphic time and space alignment algorithm module unifies the coordinate system of multi-source data through pulse neural network coding and cross-modal attention mechanism;

[0009] A blockchain trusted data pool module performs time and space mutual verification through Beidou positioning data anchoring and sensing data encryption storage;

[0010] A scene semantic decision module displays traffic risk warning and decision based on analyzing trajectory topology and point cloud semantics.

[0011] Preferably, the Beidou multi-frequency positioning module comprises:

[0012] The radio frequency front-end processing unit synchronously receives Beidou satellite B1C, B2a and B3I band signals through a multi-band microwave integrated circuit, performs signal down-conversion and low-noise amplification, and outputs an intermediate frequency signal with a signal-to-noise ratio of ≥45 dB;

[0013] The baseband signal solving unit performs carrier stripping and pseudo-code despreading on the intermediate frequency signal through a multi-channel correlator to generate raw observation data and wherein is a carrier phase observation value, is a pseudo-range observation value;

[0014] The observation quality control unit dynamically verifies the quality of the observation data through the following formula and selects centimeter-level effective data:

[0015] ;

[0016] wherein, is an observation quality factor, is a carrier phase measurement standard deviation, is a corresponding frequency band wavelength, is a pseudo-range smoothing coefficient, is an absolute value of a second-order difference of the pseudo-range.

[0017] Preferably, the heterogeneous data fusion access engine comprises:

[0018] The multi-protocol interface unit accesses multi-source data by dynamically adapting laser point cloud protocols, visual SLAM data streams and UWB positioning protocols;

[0019] The Beidou space-time reference unit establishes a space-time reference coordinate system by receiving the centimeter-level effective data output by the observation quality control unit of the Beidou multi-frequency positioning module and

[0020] The dynamic weighted fusion unit calculates the weight factors of Beidou and other perception data:

[0021] ;

[0022] wherein:

[0023] ;

[0024] is a Beidou observation quality coefficient, is a weight factor of the i-th type of perception data, is a timestamp error of the i-th type of data, is a spatial coordinate error of the i-th type of data.​

[0025] Preferably, the observation quality factor When > 0.05, the weight ratio of the enhanced laser point cloud and the UWB data is increased to 70%, and a de-bouncing filter is started to suppress transient positioning jumps.

[0026] Preferably, the neuromorphic spatio-temporal alignment algorithm module comprises:

[0027] The pulse signal coding unit generates a pulse timing feature sequence through the carrier phase observation value:

[0028] ;

[0029] Wherein, is the carrier phase change amount, is the pulse trigger threshold, is the pulse trigger time;

[0030] The cross-modal attention projection unit generates a spatial projection weight matrix by associating the pulse sequence with multi-source perception data through a multi-head attention mechanism:

[0031] ;

[0032] Wherein, is the pulse timing sequence generated by the carrier phase change trigger, is the feature vector of the multi-source perception data, is the feature dimension scaling factor;

[0033] The dynamic residual error suppression unit corrects the coordinate deviation by constraining the projection residual error.

[0034] Preferably, if the projection residual error exceeds 0.5 cm for 5 consecutive times, a re-fusion instruction is sent to the heterogeneous data fusion access engine, and the visual data access is frozen for 10 milliseconds.

[0035] Preferably, the blockchain trusted data pool module comprises the following units:

[0036] The space-time anchoring unit anchors the centimeter-level coordinates output by the Beidou multi-frequency positioning module as the genesis block of the blockchain;

[0037] The encryption evidence unit generates the space-time fingerprint of the perception data through the SHA-256 algorithm:

[0038] ;

[0039] Wherein is the aligned coordinates;

[0040] The mutual verification trigger unit detects positive coordinate deviation The weight factor is recalculated And the correction record is written into the blockchain.

[0041] Preferably, the scene semantic decision module comprises the following units:

[0042] A trajectory topology analysis unit constructs a space-time trajectory graph G=(V, E) by aligning the coordinate sequence, wherein the vertex V is a position point, and the edge E is a motion vector;

[0043] A point cloud semantic reconstruction unit identifies semantic elements such as lane lines, guardrails, and obstacles through laser point cloud data to generate a labeled grid map;

[0044] A risk decision triggering unit broadcasts a warning instruction through a V2X communication interface when a conflict between the trajectory and the semantic map is detected.

[0045] A high-precision Beidou positioning terminal device comprises:

[0046] A multi-frequency radio frequency chip set receives Beidou satellite frequency band signals through a multi-channel microwave integrated circuit and outputs carrier phase observation values;

[0047] A sensor fusion interface synchronously accesses laser radar, global shutter camera, and UWB locator data through Ethernet, MIPI-CSI, and SPI protocols;

[0048] An edge AI processor deploys a neuromorphic space-time alignment algorithm to output precise positioning coordinates under dynamic residual constraints;

[0049] A secure evidence unit generates a Beidou timestamp encrypted positioning fingerprint;

[0050] A control unit responds to semantic decision instructions and broadcasts warning messages through DSRC / LTE-V2X.

[0051] A high-precision Beidou positioning method comprises the following steps:

[0052] S1, Beidou signal processing: receiving Beidou satellite B1C, B2a, and B3I frequency band signals to generate centimeter-level carrier phase and pseudo-range original observation data, and performing dynamic quality control to select valid data;

[0053] S2, multi-source perception data fusion: dynamically accessing laser point cloud, visual SLAM, and UWB positioning data, calculating a dynamic weight factor based on space-time error, and establishing a Beidou-centered space-time reference coordinate system to realize data alignment;

[0054] S3, neuromorphic spatiotemporal alignment, through the pulse neural network coding carrier phase change event to generate pulse timing sequence, using cross-modal attention mechanism to solve spatial projection weight, output centimeter level precision alignment coordinates;

[0055] S4, blockchain spatiotemporal verification, taking Beidou centimeter level coordinates as blockchain anchor point, encrypting to generate spatiotemporal fingerprint of perception data, and triggering data re-fusion when coordinate deviation is out of standard;

[0056] S5, scene risk decision, based on the precise alignment coordinate sequence to construct motion trajectory topology, and fusing laser point cloud semantic recognition road elements to broadcast early warning instruction in real time when risk condition is detected.

[0057] The application provides a high-precision Beidou positioning terminal device and system.

[0058] 1, the application converts the physical signal layer characteristics into spatiotemporal events understandable by algorithm through the carrier phase dynamic characteristics being real-time coded into pulse timing sequence by neuromorphic module, in the tunnel scene, satellite signal fluctuation triggers high-frequency pulse, drives cross-modal attention mechanism to automatically improve laser point cloud weight, and at the same time, suppresses visual interference, reduces the error of satellite rejection scene positioning.

[0059] 2, the residual error exceeding event of neuromorphic module in the application triggers the verification mechanism of blockchain module, and drives weight factor recalculation, at the same time, fuses engine to freeze abnormal data flow, binds algorithm layer error control and security layer trusted storage, achieves double benefits of residual error exceeding event solution precision and judicial verification.

[0060] 3, the application realizes real-time collision detection of vehicle motion vector constructed by trajectory topology analysis unit through the recognized lane line and guardrail semantic boundary, cross-dimensionally correlates environmental static semantics and vehicle dynamic behavior, in the rain night scene, laser recognition guardrail position and trajectory acceleration vector fusion, avoids vehicle collision risk. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 It is an architecture diagram of a high-precision Beidou positioning system of the application;

[0062] Figure 2 It is a flowchart of a high-precision Beidou positioning method of the application. DETAILED DESCRIPTION

[0063] The technical solutions of the present application will be described clearly and completely below in combination with the drawings of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0064] Please refer to the drawings of the present application Figure 1 The embodiments of the present application provide a high-precision Beidou positioning system, comprising:

[0065] The Beidou multi-frequency positioning module generates centimeter-level raw observation data by receiving Beidou satellite B1C, B2a and B3I frequency band signals;

[0066] The heterogeneous data fusion access engine synchronously aligns the spatial structure information of the laser point cloud, the texture features of the visual SLAM and the absolute position constraints of the UWB through a dynamic weighting algorithm, so as to realize the sub-centimeter-level space-time unification of the multi-source perception data. In the satellite signal shielding scene (such as the tunnel entrance), the fusion engine automatically increases the weights of the laser and the UWB to 80%, relies on the Beidou time scale to synchronize the time stamps of the multi-source data, eliminates the positioning jump and maintains the continuous trajectory output.

[0067] The neuromorphic space-time alignment algorithm module unifies the coordinate systems of the multi-source data through the pulse neural network coding and the cross-modal attention mechanism.

[0068] The blockchain trusted data pool module anchors the Beidou positioning data and encrypts the perception data for storage and evidence, and performs space-time verification.

[0069] The scene semantic decision module displays the traffic risk warning and decision based on the analysis of the trajectory topology and the point cloud semantics.

[0070] Specifically, the Beidou multi-frequency positioning module synchronously receives the three frequency bands (B1C, B2a, B3I) of the Beidou satellite, uses the complementary characteristics of the frequency bands to solve the high-precision carrier phase and pseudo-range observation values, generates centimeter-level raw data, effectively suppresses the positioning drift caused by the multipath effect in complex urban canyon environments through multi-frequency difference, significantly improves the positioning stability in harsh scenes, realizes fast ambiguity convergence through the characteristics of the three frequency signals, greatly shortens the positioning initialization response time, and maintains the continuous positioning ability in the strong electromagnetic interference scene through intelligent frequency band switching, thereby providing a robust space-time reference for multi-source fusion.

[0071] The Beidou multi-frequency positioning module receives the B1C, B2a and B3I frequency band signals to generate centimeter-level raw observation data and provide the space-time reference origin. The heterogeneous data fusion access engine synchronously aligns the spatial structure information of the laser point cloud, the texture features of the visual SLAM and the absolute position constraints of the UWB through a dynamic weighting algorithm, so as to realize the sub-centimeter-level space-time unification of the multi-source perception data. In the satellite signal shielding scene (such as the tunnel entrance), the fusion engine automatically increases the weights of the laser and the UWB to 80%, relies on the Beidou time scale to synchronize the time stamps of the multi-source data, eliminates the positioning jump and maintains the continuous trajectory output.

[0072] The neuromorphic spatio-temporal alignment algorithm module encodes the Beidou carrier phase changes into pulse timing sequences through a pulse neural network, captures satellite signal dynamic characteristics in real time, calculates the spatial projection weight matrix of laser point cloud, visual SLAM and UWB data in combination with a cross-modal attention mechanism, and dynamically fuses multi-source perception data; in complex scenes such as tunnels, the pulse coding automatically suppresses visual data jitter, enhances laser point cloud dominant positioning, and realizes sub-centimeter level unification of multi-source coordinate systems.

[0073] The blockchain trusted data pool module anchors the Beidou centimeter-level positioning coordinates as the genesis block of the blockchain to the physical space-time reference, generates a space-time fingerprint by fusing the fine alignment coordinates through the SHA-256 algorithm, realizes the encrypted storage and tamper-proofing of perception data, and automatically triggers multi-source data re-fusion when the neuromorphic module detects that the residual error is out of standard, and writes the correction process into the distributed ledger. For example, in the supervision of port container lifting, the positioning data is realized to be traceable in the whole link, and a high-reliability space-time data base is constructed.

[0074] The scene semantic decision module analyzes the centimeter-level positioning trajectory topology output by the neuromorphic alignment, constructs a motion vector model in real time, and simultaneously fuses lane lines, guardrails and obstacle semantic labels recognized by laser point cloud. When the vehicle transverse acceleration is detected to be out of standard or the distance to the obstacle is less than the safety threshold, the V2X early warning broadcast and optical warning are triggered in milliseconds, realizing a leap from centimeter-level positioning to active safety.

[0075] The Beidou multi-frequency positioning module comprises:

[0076] The radio frequency front-end processing unit synchronously receives Beidou satellite B1C, B2a and B3I band signals through a multi-band microwave integrated circuit, performs signal down-conversion and low-noise amplification, and outputs an intermediate frequency signal with a signal-to-noise ratio of ≥45dB;

[0077] The baseband signal solving unit performs carrier stripping and pseudo-code despreading on the intermediate frequency signal through a multi-channel correlator to generate raw observation data and wherein is a carrier phase observation value, is a pseudo-range observation value;

[0078] The observation quality control unit dynamically verifies the quality of the observation data through the following formula and selects the centimeter-level effective data:

[0079] ;

[0080] wherein, is an observation quality factor, is a carrier phase measurement standard deviation, is a wavelength corresponding to the frequency band, The pseudo-range smoothing coefficient is The pseudo-range second-order difference absolute value is

[0081] Specifically, the radio frequency front-end processing unit synchronously receives the Beidou three-frequency signal (B1C / B2a / B3I) through a multi-frequency microwave integrated circuit, performs signal down-conversion and low-noise amplification, ensures that the signal-to-noise ratio of the output intermediate frequency signal is greater than or equal to 45 dB, and in the strong interference scene (1.4 GHz frequency band) of the 5G base station, combined with the adaptive band-stop filter to suppress the in-band noise, provides a pure signal source for centimeter-level observation.

[0082] The baseband signal solving unit performs carrier stripping and pseudo-code decoding on the intermediate frequency signal through a multi-channel correlator to generate carrier phase observation values (millimeter-level precision) and pseudo-range observation values (centimeter-level precision), solves ionospheric delay errors through dual-frequency combination, greatly improves the ambiguity fixing efficiency of dynamic positioning, and supports high-frequency real-time centimeter-level raw data output.

[0083] The observation quality control unit filters effective data based on a dynamic verification formula, so as to ensure that the output data meet the centimeter-level positioning reliability standard through a multi-path error identification and filtering mechanism.

[0084] The heterogeneous data fusion access engine includes:

[0085] The multi-protocol interface unit accesses multi-source data by dynamically adapting laser point cloud protocols, visual SLAM data streams, and UWB positioning protocols;

[0086] The Beidou space-time reference unit establishes a space-time reference coordinate system by receiving the centimeter-level effective data output by the observation quality control unit of the Beidou multi-frequency positioning module and .

[0087] The dynamic weighted fusion unit calculates the weight factors of Beidou and other perception data:

[0088] ;

[0089] Wherein:

[0090] ;

[0091] The Beidou observation quality coefficient is The weight factor of the i-th type of perception data is The timestamp error of the i-th type of data is The spatial coordinate error of the i-th type of data is

[0092] Specifically, the multi-protocol interface unit realizes millisecond-level access of multi-source heterogeneous data by dynamically adapting laser point cloud protocol (EtherCAT), visual SLAM data stream (MIPI-CSI2), and UWB positioning protocol (IEEE 802.15.4a), synchronously analyzes Velodyne laser radar point cloud, ZED visual camera feature points, and Decawave UWB ranging signals in the terminal, eliminates the communication delay difference between devices (timestamp alignment error <1 ms), and provides standardized data input for the fusion engine.

[0093] Beidou space-time reference unit The output of the centimeter-level effective data (carrier phase observation value , pseudo-range observation value ) is the origin, a unified space-time reference is established through WGS-84 coordinate system conversion, in the satellite signal interruption scene (such as a tunnel), relying on Beidou timing to maintain the time reference, combined with UWB absolute position constraint, the spatial offset error of multi-source data is reduced.

[0094] The dynamic weighted fusion unit calculates the weight in real time based on the formula for the timestamp error, for the spatial error, and Beidou is the Beidou quality coefficient), when the rain causes the spatial error of the visual SLAM to increase, the weight is dynamically adjusted to ensure the continuity of the fusion output trajectory.

[0095] Observation quality factor >0.05, the weight ratio of laser point cloud and UWB data is increased to 70%, and a debouncing filter is started to suppress instantaneous positioning jumps.

[0096] Specifically, when the Beidou observation quality factor >0.05, the fusion weight of laser point cloud and UWB data is automatically increased to 70% and the debouncing filter is activated, in the satellite signal degradation scene (such as a tunnel / strong electromagnetic interference), to suppress the instantaneous positioning jumps caused by multipath noise and maintain the centimeter-level trajectory continuity.

[0097] The neuromorphic space-time alignment algorithm module includes:

[0098] The pulse signal coding unit generates a pulse timing feature sequence through the carrier phase observation value:

[0099] ;

[0100] wherein, is the carrier phase change amount, is the pulse trigger threshold, is the pulse trigger time;

[0101] The cross-modal attention projection unit correlates the pulse sequence and multi-source perception data through a multi-head attention mechanism to generate a spatial projection weight matrix.

[0102]

[0103] wherein, is a pulse timing sequence generated by carrier phase change triggering, is a feature vector of multi-source perception data, is a feature dimension scaling factor;

[0104] The dynamic residual suppression unit corrects coordinate deviation by constraining projection residual.

[0105] Specifically, the pulse signal encoding unit calculates the Beidou carrier phase change amount in real time When the change amount exceeds the adaptive pulse trigger threshold (θ=k·σnoise), a pulse timing sequence is generated Accurately capture satellite signal dynamic characteristics, for example, in the vehicle rapid acceleration scenario, the carrier phase mutation triggers high-frequency pulses, significantly improving the response sensitivity to motion state changes, providing event-driven neuromorphic coding basis for multi-source alignment;

[0106] The cross-modal attention projection unit

[0107] Based on the multi-head attention mechanism, the pulse timing sequence is associated with the perception feature vector of laser point cloud, visual SLAM, and UWB to calculate the spatial projection weight matrix In the visual rain and fog interference scene, automatically suppress the visual weight to 0.1 and increase the laser point cloud weight to 0.8, realize intelligent fusion of cross-modal features, and eliminate the spatial deviation of multi-source data;

[0108] The dynamic residual suppression unit corrects the coordinate output in real time by constraining the projection residual of Beidou. When the residual exceeds the standard due to continuous curves of overpass, it immediately sends a re-fusion instruction to the fusion engine and freezes the abnormal data stream, suppresses the positioning trajectory fluctuation within the internal, and solves the cumulative error diffusion problem in complex scenes.

[0109] If the projection residual exceeds 0.5 cm for 5 consecutive times, send a re-fusion instruction to the heterogeneous data fusion access engine, and freeze the visual data access for 10 milliseconds.

[0110] Specifically, by dynamically monitoring the projection residual exceeding the 0.5cm tolerance threshold for 5 consecutive times, a re-fusion instruction is sent to the heterogeneous data fusion access engine, and the visual data access is frozen for 10 milliseconds; in the scene where strong electromagnetic interference or extreme weather causes visual data distortion, the abnormal data pollution is quickly blocked, triggering real-time weight reset (laser point cloud weight is increased to 85%), effectively solving the cumulative error diffusion problem, and ensuring the robustness of the centimeter-level positioning link.

[0111] The blockchain trusted data pool module includes the following units:

[0112] The space-time anchoring unit anchors the centimeter-level coordinates output by the Beidou multi-frequency positioning module to the blockchain, and the space-time anchoring unit includes the following units: As the genesis block of the blockchain;

[0113] The encryption storage unit generates a space-time fingerprint of the perception data through the SHA-256 algorithm:

[0114] ;

[0115] Wherein is the aligned coordinates;

[0116] The mutual verification triggering unit recalculates the weight factor and writes the correction record into the blockchain when detecting positive coordinate deviation .

[0117] Specifically, the space-time anchoring unit anchors the Beidou centimeter-level positioning coordinates as the genesis block hash root of the blockchain, establishing an unalterable physical space-time origin. In the port container scheduling scene, this unit anchors the crane positioning coordinates to the blockchain, realizing legal-level trusted storage of the operation position;

[0118] The encryption storage unit generates a space-time fingerprint by fusing the fine alignment coordinates and the Beidou original coordinates through the SHA-256 algorithm, that is, when the vehicle changes lanes on the highway, the trajectory fingerprint is stored in real time to the distributed ledger, and a judicial traceability technology closed loop is constructed;

[0119] The mutual verification triggering unit automatically triggers weight factor recalculation when detecting that the residual of the neuromorphic alignment output is over-standard, and the residual abnormal point coordinates and the correction record are encrypted and written into the blockchain.

[0120] The scene semantic decision module includes the following units:

[0121] The trajectory topology analysis unit constructs a space-time trajectory graph G=(V,E) by aligning the coordinate sequence, where the vertex V is the position point and the edge E is the motion vector;

[0122] The point cloud semantic reconstruction unit identifies semantic elements such as lane lines, guardrails, and obstacles through laser point cloud data, and generates a labeled grid map.

[0123] A risk decision trigger unit broadcasts a warning instruction through a V2X communication interface when detecting that the trajectory conflicts with the semantic map.

[0124] Specifically, the trajectory topology analysis unit constructs a space-time trajectory graph through a high-precision coordinate sequence, real-time solves vehicle kinematic characteristics, accurately captures risk characteristics such as lateral acceleration mutation and heading deviation, dynamically generates a trajectory topology model in a complex road scene, provides millisecond-level motion state analysis for risk decision, and significantly improves the active protection capability of driving safety;

[0125] The point cloud semantic reconstruction unit constructs a grid-based environment map with semantic labels by analyzing the spatial distribution characteristics of laser point clouds (such as identifying continuous linear point clouds as lane lines, height mutation point clouds as guardrails, and moving cluster point clouds as obstacles). In low-visibility scenarios such as rain, fog, and night, it breaks through the perception limitations of traditional visual solutions, provides robust road structure semantic expression for the risk decision module, realizes accurate cognition of all-weather environmental elements, and lays the foundation for active safety decision-making;

[0126] The risk decision trigger unit compares the motion vector output by the trajectory topology analysis unit with the environment semantic map generated by the point cloud semantic reconstruction unit in real time. When detecting that the vehicle trajectory conflicts with the road structure (such as deviating from the lane or approaching obstacles), it triggers a multi-level response protocol in milliseconds, broadcasts a standardized warning instruction through a V2X communication interface, and synchronously links the vehicle control system to execute avoidance actions, realizing active safety closed-loop protection in complex traffic scenarios.

[0127] A high-precision Beidou positioning terminal device, comprising:

[0128] A multi-frequency radio frequency chip set receives Beidou satellite frequency band signals through a multi-channel microwave integrated circuit and outputs carrier phase observation values;

[0129] A sensor fusion interface synchronously accesses laser radar, global shutter camera, and UWB locator data through Ethernet, MIPI-CSI, and SPI protocols;

[0130] An edge AI processor deploys a neuromorphic space-time alignment algorithm and outputs precise positioning coordinates under dynamic residual constraints;

[0131] A secure storage unit generates a Beidou timestamp encrypted positioning fingerprint;

[0132] A control unit responds to semantic decision instructions and broadcasts warning messages through DSRC / LTE-V2X.

[0133] Specifically, the terminal as a hardware carrier realizes high-sensitivity reception of Beidou three-frequency signals through a multi-frequency radio frequency chipset, a sensor fusion interface synchronously accesses heterogeneous data of a laser radar, a visual camera and UWB, and an edge AI processor solves centimeter-level accurate positioning coordinates in real time based on a neuromorphic algorithm; a secure storage unit generates a Beidou timing encrypted space-time fingerprint to build a trusted data base, and a control unit triggers a V2X early warning broadcast in milliseconds according to a semantic decision result, thereby forming a hardware-level closed loop from satellite signal reception to active safety decision.

[0134] Please refer to the attached Figure 2 , a high-precision Beidou positioning method, comprising the following steps:

[0135] S1, Beidou signal processing, through receiving Beidou satellite B1C, B2a and B3I frequency band signals, generating centimeter-level carrier phase and pseudo-range original observation data, and performing dynamic quality control to select effective data;

[0136] S2, multi-source perception data fusion, dynamically accessing laser point cloud, visual SLAM and UWB positioning data, calculating dynamic weight factors based on space-time error, establishing a Beidou-centered space-time reference coordinate system to realize data alignment;

[0137] S3, neuromorphic space-time alignment, encoding carrier phase change events through a pulse neural network to generate a pulse time sequence, using a cross-modal attention mechanism to solve spatial projection weights, and outputting centimeter-level accurate alignment coordinates;

[0138] S4, blockchain space-time verification, taking Beidou centimeter-level coordinates as a blockchain anchor point, and generating a space-time fingerprint of perception data for storage, and triggering data re-fusion when the coordinate deviation exceeds the standard;

[0139] S5, scene risk decision, constructing a motion trajectory topology based on the accurate alignment coordinate sequence, fusing laser point cloud semantic recognition of road elements, and broadcasting a warning instruction in real time when a risk condition is detected.

[0140] Specifically, S1 generates centimeter-level carrier phase and pseudo-range original data by receiving Beidou three-frequency signals, performs dynamic quality control for screening, provides a high-reliability space-time reference source for multi-source fusion, and solves the industry pain point that satellite signals are easily disturbed in complex environments.

[0141] S2 dynamically accesses laser / visual / UWB multi-source data, calculates weight factors based on space-time error to establish a Beidou-centered coordinate system, realizes sub-centimeter-level space-time alignment of heterogeneous perception data, and solves the problem of non-uniformity of multi-source reference.

[0142] S3 generates pulse sequences by encoding carrier phase dynamic events with pulse neural network, calculates multi-source spatial projection weights across modal attention mechanism, automatically suppresses abnormal data in tunnel / rain and fog scenes, and outputs precise alignment coordinates to ensure centimeter-level precision continuity in complex environments.

[0143] S4 generates time-space fingerprints by encrypting Beidou coordinates as blockchain anchors, triggers re-fusion and on-chain recording of residual anomalies, builds a judicial-level trusted traceability base, and breaks through the bottleneck of easily tampered positioning data.

[0144] S5 constructs a motion trajectory topology model based on precise alignment coordinates, fuses laser point cloud semantic recognition of road elements, detects trajectory deviation / obstacle intrusion risk in real time and broadcasts warning instructions in stages, and realizes a closed-loop leap from positioning to active safety.

[0145] Although embodiments of the present application have been shown and described, it will be understood by those having ordinary skill in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-precision BeiDou positioning system, characterized in that, include: The BeiDou multi-frequency positioning module receives signals from the B1C, B2a, and B3I frequency bands of BeiDou satellites and generates centimeter-level raw observation data. The heterogeneous data fusion access engine dynamically accesses laser point cloud, visual SLAM and UWB positioning data to align multi-source sensing data in time and space. The neuromorphic spatiotemporal alignment algorithm module unifies the coordinate systems of multi-source data through spiking neural network encoding and cross-modal attention mechanism. The blockchain trusted data pool module uses the centimeter-level coordinates output by the BeiDou multi-frequency positioning module as the blockchain genesis block and generates a spatiotemporal fingerprint of the sensing data using the SHA-256 algorithm. When a coordinate deviation is detected... When necessary, the weighting factors are recalculated, and the correction record is written to the blockchain; The scene semantic decision-making module constructs a spatiotemporal trajectory map by aligning coordinate sequences and identifies semantic elements of lane lines, guardrails, and obstacles using laser point cloud data, broadcasting warning commands for traffic risk warnings and decisions.

2. The high-precision BeiDou positioning system according to claim 1, characterized in that, The BeiDou multi-frequency positioning module includes: The radio frequency front-end processing unit synchronously receives signals from the B1C, B2a and B3I bands of Beidou satellites through a multi-band microwave integrated circuit, performs signal down-conversion and low-noise amplification, and outputs an intermediate frequency signal with a signal-to-noise ratio of ≥45dB. The baseband signal processing unit performs carrier stripping and pseudo-code despreading on the intermediate frequency signal through a multi-channel correlator to generate raw observation data. and ,in For carrier phase observations, These are pseudorange observations; Observation quality control unit: Dynamically verifies the quality of observation data and filters out valid data at the centimeter level using the following formula: ; in, To observe the quality factor, Standard deviation of carrier phase measurement For the corresponding frequency band wavelength, The pseudo-range smoothing coefficient, It is the absolute value of the second-order difference of the pseudo-distance.

3. The high-precision BeiDou positioning system according to claim 1, characterized in that, The heterogeneous data fusion access engine includes: The multi-protocol interface unit can access multi-source data by dynamically adapting to laser point cloud protocols, visual SLAM data streams, and UWB positioning protocols; The BeiDou spatiotemporal reference unit receives centimeter-level effective data output from the observation quality control unit of the BeiDou multi-frequency positioning module. and Establish a spatiotemporal reference coordinate system. To observe the quality factor, These are pseudorange observations; The dynamic weighted fusion unit calculates the weighting factors of BeiDou data and other sensing data. ; in: ; The quality coefficient of BeiDou observation. For the first Weighting factors for class-aware data For the first Timestamp error of data type For the first Spatial coordinate error of class data.

4. A high-precision BeiDou positioning system according to claim 3, characterized in that, The observation quality factor At the same time, the weight ratio of laser point cloud and UWB data is increased to 70%, and the jitter filter is activated to suppress instantaneous positioning jumps.

5. A high-precision BeiDou positioning system according to claim 1, characterized in that, The neuromorphic spatiotemporal alignment algorithm module includes: The pulse signal encoding unit generates a pulse timing feature sequence based on carrier phase observations. ; in, This is the carrier phase change. The pulse trigger threshold, This refers to the pulse trigger moment; The cross-modal attention projection unit associates pulse sequences with multi-source sensor data through a multi-head attention mechanism to generate a spatial projection weight matrix: ; in, It is a pulse timing sequence, generated by carrier phase changes. It is a feature vector of multi-source sensing data. It is the feature dimension scaling factor; The dynamic residual suppression unit corrects coordinate deviations by constraining the projected residuals.

6. A high-precision BeiDou positioning system according to claim 5, characterized in that, If the projection residual exceeds 0.5cm five times consecutively, a re-fusion command is sent to the heterogeneous data fusion access engine, and the visual data access is frozen for 10 milliseconds.

7. A high-precision BeiDou positioning system according to claim 1, characterized in that, The blockchain trusted data pool module includes the following units: The spatiotemporal anchoring unit uses centimeter-level coordinates output by the BeiDou multi-frequency positioning module. As the genesis block of the blockchain; The encrypted evidence storage unit generates a spatiotemporal fingerprint of the perceived data using the SHA-256 algorithm. ; in To align coordinates; The mutual verification trigger unit detects a positive coordinate deviation. When recalculating the weighting factors And the correction record will be written to the blockchain.

8. A high-precision BeiDou positioning system according to claim 1, characterized in that, The scene semantic decision-making module includes the following units: The trajectory topology parsing unit constructs a spatiotemporal trajectory graph G=(V,E) by aligning coordinate sequences, where vertex V is the position point and edge E is the motion vector; The point cloud semantic reconstruction unit uses laser point cloud data to identify semantic elements of lane lines, guardrails, and obstacles, and generates a labeled raster map. The risk decision triggering unit broadcasts a warning command through the V2X communication interface when it detects a conflict between the trajectory and the semantic map.

9. A high-precision BeiDou positioning terminal device, characterized in that, A high-precision BeiDou positioning system according to any one of claims 1-8, comprising: The multi-frequency radio frequency chipset receives BeiDou satellite frequency band signals through a multi-channel microwave integrated circuit and outputs carrier phase observation values. The sensor fusion interface synchronously accesses data from LiDAR, global shutter camera, and UWB locator via Ethernet, MIPI-CSI, and SPI protocols. An edge AI processor deploys a neuromorphic spatiotemporal alignment algorithm to output precise positioning coordinates under dynamic residual constraints; The secure evidence storage unit generates a location fingerprint encrypted with a BeiDou timestamp; The control unit responds to semantic decision commands and broadcasts warning messages via DSRC / LTE-V2X.

10. A high-precision BeiDou positioning method, characterized in that, A high-precision BeiDou positioning system according to any one of claims 1-8, comprising the following steps: S1. BeiDou signal processing: By receiving signals from BeiDou satellites in the B1C, B2a and B3I frequency bands, it generates centimeter-level carrier phase and pseudorange raw observation data, and performs dynamic quality control to filter valid data. S2. Multi-source sensing data fusion, dynamically accessing laser point cloud, visual SLAM and UWB positioning data, calculating dynamic weighting factors based on spatiotemporal error, and establishing a spatiotemporal reference coordinate system centered on BeiDou to achieve data alignment. S3, Neuromorphic spatiotemporal alignment, generates pulse time sequence by encoding carrier phase change events through a spiking neural network, solves spatial projection weights using a cross-modal attention mechanism, and outputs centimeter-level precise alignment coordinates; S4. Blockchain spatiotemporal mutual verification uses BeiDou centimeter-level coordinates as blockchain anchor points to encrypt and generate spatiotemporal fingerprint evidence of sensing data. When the coordinate deviation exceeds the standard, data re-fusion is triggered. S5, Scene Risk Decision-Making: Constructs motion trajectory topology based on precisely aligned coordinate sequences, integrates laser point cloud semantic recognition of road elements, and broadcasts early warning instructions in real time when risk conditions are detected.

Images