CN120282264A

CN120282264A - Positioning system and method for internal and external motorcades of port

Info

Publication number: CN120282264A
Application number: CN202510748704.5A
Authority: CN
Inventors: 顾嘉俊; 邱长伍
Original assignee: Mengshi Technology Suzhou Co ltd
Current assignee: Mengshi Technology Suzhou Co ltd
Priority date: 2025-06-06
Filing date: 2025-06-06
Publication date: 2025-07-08
Anticipated expiration: 2045-06-06
Also published as: CN120282264B

Abstract

The invention discloses a system and a method for positioning internal and external motorcades in a port, and relates to the technical field of intelligent positioning. The method is used for solving the problems of frequent positioning blind areas, sensitive dynamic interference, uncontrollable error accumulation and low anchor point resource scheduling efficiency in complex harbor scenes. And dynamically calculating anchor point coverage through the attenuation characteristic of the metal environment signal, generating blind area information by combining with vehicle track clustering, and optimizing signal resource allocation. And performing ground texture gray entropy segmentation and cross-frame Bayesian association on the blind area, eliminating interference to generate synchronous pose data, and improving visual positioning robustness. And constructing an error propagation chain model based on the inter-vehicle communication topology, integrating the multiple-vehicle confidence weight correction accumulated errors, and enhancing the positioning consistency of the vehicle fleet. And finally, analyzing the port scheduling plan and the vehicle motion state, constructing a space-time prediction model to dynamically generate an anchor point instruction, and realizing the on-demand scheduling of anchor point resources. Remarkably improves the positioning precision and the resource utilization efficiency of the port vehicle, and provides reliable support for port automation operation.

Description

Positioning system and method for internal and external motorcades of port

Technical Field

The invention relates to the technical field of intelligent positioning, in particular to a system and a method for positioning internal and external motorcades in a port.

Background

Along with the continuous improvement of the automation and the intellectualization level of the global port, the container terminal gradually changes to an operation mode of paperless dispatch, unmanned carrying and informatization collaboration. The high-frequency cooperative operation network is formed among the collection cards, the stacking machines, the Automatic Guided Vehicle (AGV) operation vehicles in the port and logistics trucks at the periphery of the port, and higher requirements are set for real-time positioning accuracy, system robustness and dispatching response capability of the vehicles. In a typical harbor operation environment, due to dense metal equipment and serious signal shielding, the traditional positioning mode which is singly dependent on satellite navigation has poor stability and large error floating, and is difficult to support complex dispatching tasks and automatic operation flows. In order to ensure the operation safety and improve the throughput efficiency, it is highly desirable to establish a high-precision multi-vehicle co-location system which is oriented to port scenes and has dynamic self-adaptation capability.

The existing harbor vehicle positioning method mainly comprises three types, namely an absolute positioning system based on a Global Navigation Satellite System (GNSS), an auxiliary positioning system based on ground deployment Ultra Wideband (UWB) signal positioning, radio Frequency Identification (RFID) tag identification or anchor point identification such as magnetic nails and the like, and a feature perception positioning system based on vision and laser radar. However, the practical application of these methods has the following substantial technical drawbacks that firstly, the anchor point positioning scheme adopts a fixed deployment mode, and lacks a mechanism for real-time cooperation with the vehicle operation path, so that the anchor point signal utilization rate is low, and a signal blind area is easy to form in a dynamic operation area. Secondly, the visual positioning method generally adopts single-frame image feature extraction, is easily influenced by interference factors such as dynamic shielding objects, illumination changes and the like, and cannot stably output continuous pose estimation results. Thirdly, a centralized processing architecture is generally adopted in the multi-vehicle co-location system, the difference of communication relations and location uncertainty among nodes in a vehicle team is not considered, and errors are easy to accumulate and spread in the system. Fourthly, the anchor point activation strategy is mostly periodically and pollingly triggered, is not linked with an operation plan in a port scheduling system in real time, cannot realize anchor point coverage prediction based on future motion trend of a vehicle, has delayed scheduling control response, and affects the overall cooperative performance of the system.

Disclosure of Invention

Aiming at the defects of the prior art, the invention provides a system and a method for positioning internal and external motorcades in a port, and solves the problems of the background technology.

The system comprises a dynamic anchor point collaborative construction module, an anti-interference environment characteristic extraction module, a vehicle team collaborative error suppression module and an anchor point space-time prediction module, wherein the dynamic anchor point collaborative construction module is used for acquiring dynamic position data of port mobile equipment in real time, dynamically calculating anchor point signal coverage according to signal propagation attenuation characteristics under a metal environment and generating anchor point activation instructions and anchor point coverage blind area information by combining vehicle historical track clustering analysis, the anti-interference environment characteristic extraction module is used for receiving the anchor point coverage blind area information, carrying out local gray entropy segmentation on ground images acquired by a vehicle-mounted camera, screening high-distinction texture areas, generating position estimation data aligned with anchor point signal space time through cross-frame Bayesian probability association, the vehicle team collaborative error suppression module is used for receiving the position estimation data and vehicle inertial navigation data, constructing an error propagation chain model by combining with vehicle-to-vehicle communication topological relation, correcting accumulated positioning errors by a multi-vehicle confidence weight fusion strategy, and the anchor point space-time prediction module is used for analyzing operation planning of mobile equipment in a scheduling system, carrying out local gray entropy segmentation on ground images acquired by a vehicle camera, screening high-distinction texture areas, generating position estimation data aligned with anchor point space-time position estimation data according to the anchor point spatial prediction data, and generating a vehicle position prediction matrix according to the position prediction model with the position prediction state of the motion state of the mobile equipment in the vehicle has priority, and the position prediction module can reach the position prediction matrix.

Further, the method comprises the specific processes of analyzing the dynamic position data of the port mobile equipment in real time, obtaining real-time coordinates and moving directions of a portal crane boom and temporary traffic equipment, adjusting attenuation coefficients of a signal propagation model in a segmented mode according to the distribution density of metal obstacles in a container stacking area, generating a non-uniform attenuation gradient, calculating signal effective coverage boundaries from anchor point to anchor point according to the corrected attenuation gradient, generating a vector map described by polygon vertex coordinates, matching the real-time position of a vehicle with the vector map in a geometric relationship mode, marking the area with the signal strength lower than a positioning threshold as a coverage blind area, and outputting a blind area vertex coordinate list.

The method comprises the specific processes of collecting vehicle history track data, identifying a high-frequency passing area comprising a container transfer area and a gate channel through a density clustering algorithm, generating an anchor point activation priority order according to a clustering result, preferentially activating a mobile device anchor point in the high-frequency area, matching the coverage blind point coordinates with real-time positions of vehicles, activating adjacent mobile device anchor points according to the blind point vertex coordinates when the vehicles drive into the blind point boundaries, prolonging the working time of the adjacent mobile device anchor points, updating an anchor point activation time sequence table according to the real-time positions of the vehicles, and adjusting signal coverage areas.

Further, the specific process of screening the high-resolution texture region comprises dividing the ground image acquired by the vehicle-mounted camera into a plurality of square local blocks in the coverage blind area, calculating the local gray entropy value of each block, screening the blocks with the gray entropy value higher than the environment background as candidate texture regions, comparing the entropy value changes of the same blocks in the adjacent frame images, and eliminating the abnormal high-entropy blocks caused by instantaneous reflection or projection; and performing morphological closing operation on the candidate texture region, filling the fracture region, and generating a continuous positioning reference mask.

The specific process of generating pose estimation data aligned with anchor point signal time and space includes the steps of performing time and space calibration on stable texture features in continuous multi-frame images, recording feature point motion tracks, constructing a cross-frame feature correlation probability matrix, calculating correlation probability of current frame features and historical frames according to a Bayesian inference model, eliminating the features if the probability is lower than a dynamic interference threshold value, aligning the screened features with anchor point signal time stamps, solving vehicle poses through a perspective transformation matrix, fusing multi-frame pose estimation results, and generating smoothed vehicle track data.

Further, the construction logic for receiving pose estimation data and vehicle inertial navigation data and constructing an error propagation chain model by combining the communication topological relation among vehicles is as follows. According to the consistency of vehicle motion direction, the transmission attenuation coefficient of the error along the communication link is calculated to generate an error propagation chain matrix, and when the vehicle position or the communication topology changes, the error propagation weight and the attenuation coefficient are recalculated.

The method comprises the specific processes of correcting accumulated positioning errors through a multi-vehicle confidence weight fusion strategy, namely dynamically distributing confidence weights according to the alignment degree and historical positioning stability of vehicle pose estimation data and anchor point signals, reducing abnormal vehicle weights and triggering local texture matching if the adjacent vehicle pose difference exceeds a set threshold value, fusing multi-vehicle pose data through a weighted average algorithm based on an error propagation chain matrix to restrain single vehicle errors, feeding back fused poses to each vehicle node, and updating an inertial navigation initial state.

Further, analyzing an operation plan of the mobile device in the port scheduling system, and constructing a space-time reachability prediction model by combining the motion state of the corrected vehicle pose; according to the corrected pose data, a vehicle kinematic model is constructed, a space-time region reachable by the vehicle in a future period is predicted, intersection operation is carried out on a vehicle predicted track and space-time distribution of anchor points of mobile equipment, a space-time reachability probability distribution map is generated, an anchor point activation priority list is generated according to space-time density of the intersection region, and high-probability touch anchor points are marked.

The method comprises the specific processes of predicting a time-space distribution matrix of future reachable anchors, generating a control instruction according to the activation priority of the anchors in the matrix, and feeding back the control instruction to a dynamic anchor cooperative construction module, wherein the time-space reachability probability distribution map is quantized into the matrix, the dimension of the matrix is a time-space-anchor identifier, the order of activating the priority of the anchors is generated according to the probability value sequence of space-time nodes in the matrix, the high-probability reachable anchors are preferentially awakened, the priority instruction is fed back to the dynamic anchor cooperative construction module through a port wireless private network, anchor awakening or dormancy operation is triggered, anchor activating effects are monitored in real time, and if a vehicle does not reach the anchors according to the prediction, the priority instruction is dynamically adjusted, and the matrix is rebuilt.

A method for locating the internal and external vehicle teams in a port includes such steps as S1, obtaining the dynamic position data of mobile equipment in the port in real time, dynamically calculating the coverage range of anchor point signals according to the signal propagation attenuation characteristics in metal environment, generating anchor point activating instruction and anchor point coverage blind area information by combining with the vehicle history track clustering analysis, S2, receiving the anchor point coverage blind area information, dividing the ground image collected by the vehicle-mounted camera by local grey entropy, screening high-discrimination texture regions, eliminating dynamic interference by cross-frame Bayesian probability association, generating pose estimation data aligned with the anchor point signals in time-space mode, S3, constructing error propagation chain model by combining with the communication topological relation between vehicles, correcting accumulated locating errors by multi-vehicle confidence weight fusion strategy, S4, analyzing the operation plan of mobile equipment in the port dispatching system, constructing time-space accessibility prediction model by combining with the motion state of corrected vehicle pose, predicting the time-space distribution matrix of future touchable anchor points, generating control instruction according to the anchor point activating priority in the matrix, and feeding back to S1.

The invention has the following beneficial effects:

(1) According to the system for positioning the internal and external motorcades of the port, the coverage area of the anchor point signal is dynamically calculated according to the signal propagation attenuation characteristic under the metal environment through the dynamic anchor point cooperative construction module, the anchor point activating instruction and the blind area information are generated by combining the vehicle history track cluster analysis, the problem of coverage blind areas caused by signal multipath reflection under the metal dense scene of the port is effectively solved, and the utilization rate and positioning continuity of anchor point resources are improved. Through the anti-interference environment feature extraction module, local gray entropy segmentation and cross-frame Bayesian probability association are carried out on the ground image in the coverage blind area, high-distinction texture features are screened out, dynamic interference (such as container projection and accumulated water reflection) is eliminated, the space-time alignment precision of visual positioning and anchor point signals is remarkably improved, and the robustness of positioning data in a complex environment is ensured.

(2) The method for positioning the internal and external motorcades in the port builds an error propagation chain model based on the dynamic communication topological relation among vehicles, corrects the accumulated positioning errors by combining a multi-vehicle confidence weight fusion strategy, breaks through the bottleneck of long-term error accumulation of single-vehicle inertial navigation, and improves the overall positioning stability of the motorcades. Analyzing the operation plan of the port scheduling system, constructing a space-time reachability prediction model by combining the motion state of the vehicle, dynamically generating an anchor point activation priority order, feeding back the order to an anchor point cooperative construction module in a closed loop, realizing the on-demand scheduling and predictive response of anchor point resources, greatly reducing the dependence of a positioning system on a fixed infrastructure, and simultaneously improving the scheduling efficiency and the operation safety of a port temporary motorcade.

Of course, it is not necessary for any one product to practice the invention to achieve all of the advantages set forth above at the same time.

Drawings

FIG. 1 is a flow chart of the system for locating internal and external motorcades in a port according to the present invention.

FIG. 2 is a flow chart of the method for locating the internal and external motorcades in the port according to the invention.

Detailed Description

According to the embodiment of the application, through the system and the method for positioning the internal and external vehicle teams in the port, the problems of frequent vehicle positioning blind areas, sensitive dynamic interference, uncontrollable error accumulation and low anchor point resource scheduling efficiency in the complex port environment are solved.

The scheme in the embodiment of the application has the following overall thought:

Dynamic anchor point cooperative construction, namely dynamically generating anchor point activating instructions and coverage blind area information by acquiring dynamic position data of port mobile equipment in real time and combining a metal environment signal attenuation correction model and vehicle history track cluster analysis, so as to solve the problems of high deployment cost of a fixed anchor point and randomness of a signal coverage blind area.

And extracting anti-interference environmental characteristics, namely screening ground high-resolution textures based on local gray entropy segmentation in a coverage blind area, eliminating instantaneous interference (such as container projection and accumulated water reflection) through cross-frame Bayesian probability association, generating pose data aligned with anchor point signals in time-space, and breaking through the dependence of visual positioning on artificial marks.

And the motorcade cooperative error suppression is realized by constructing an error propagation chain model based on a dynamic communication topological relation among vehicles, correcting inertial navigation accumulated errors through a multi-vehicle confidence weight fusion strategy, solving the problem of long-term positioning drift of a single vehicle and improving the overall positioning consistency of the motorcade.

Analyzing the operation plan of the port scheduling system, constructing a space-time reachability prediction model by combining the motion state of the vehicle, dynamically generating an anchor activating priority instruction and feeding back the instruction to an anchor coordination module, realizing the on-demand scheduling and predictive response of anchor resources, and reducing the hardware dependent cost.

Referring to fig. 1, the embodiment of the invention provides a technical scheme, which comprises a dynamic anchor point collaborative construction module, an anti-interference environment characteristic extraction module, a vehicle team collaborative error suppression module and an anchor point space-time prediction module, wherein the dynamic anchor point collaborative construction module is used for acquiring dynamic position data of port mobile equipment in real time, dynamically calculating anchor point signal coverage according to signal propagation attenuation characteristics under a metal environment and generating anchor point activation instructions and anchor point coverage blind area information by combining vehicle historical track clustering analysis, the anti-interference environment characteristic extraction module is used for receiving the anchor point coverage blind area information, carrying out local gray entropy segmentation on ground images acquired by a vehicle-mounted camera, screening high-resolution texture areas, eliminating dynamic interference through cross-frame Bayesian probability association, generating pose estimation data aligned with anchor point signal space-time, the vehicle team collaborative error suppression module is used for receiving pose estimation data and vehicle inertial navigation data, constructing an error propagation chain model by combining with a vehicle inter-vehicle communication topological relation, correcting positioning errors through a multi-vehicle confidence weight fusion strategy, and the anchor point prediction module is used for constructing an operation plan of mobile equipment in the port mobile equipment by combining with a corrected motion state of the vehicle position, and can reach a future anchor point prediction model with a priority and reach a prediction matrix to be activated by the dynamic state prediction matrix.

In the implementation scheme, the dynamic anchor point cooperative construction module is used for dynamically establishing an anchor point network according to the real-time position change of mobile equipment (such as AGVs, collector cards and the like) in the port. The method aims to improve the space distribution rationality and dynamic adaptability of the positioning signals and avoid signal redundancy and dead zones. Dynamic position data refers to position information (which can be obtained by GNSS, inertial navigation or combined navigation) uploaded by mobile equipment in a port in real time. Signal propagation attenuation characteristics, namely obvious attenuation and multipath effect can appear in electromagnetic wave propagation under a metal dense environment. This characteristic is considered in the present invention for dynamically estimating the effective signal coverage of the anchor point. Anchor signal coverage refers to the area where a single anchor device can stably provide positioning service under the current environmental conditions. And (3) carrying out vehicle historical track clustering analysis, namely extracting a conventional path and a high-frequency operation area by carrying out track clustering on the historical positioning tracks, and providing a statistical basis for anchor point layout and activation decision. Anchor activation instructions for indicating which anchors should be activated for the current or future time period to provide location services and blind zone information for marking areas where the current anchor signal is not reachable. The anti-interference environment characteristic extraction module is used for carrying out auxiliary positioning by means of the vehicle-mounted visual perception system in the blind area covered by the anchor point signal, so that the system can still estimate the vehicle pose stably under the unavailable anchor point scene. And the local gray entropy segmentation is used for extracting texture areas with rich information and high discrimination based on local entropy values of gray distribution in the image, and is beneficial to improving the stability and uniqueness of image characteristics. Gray entropy, which is the degree of confusion (information entropy) reflecting gray distribution in an image, is used for judging the complexity of a region. The high-resolution texture region is a part with obvious boundary, clear structure and low repeatability in the image, and is suitable for feature matching and pose estimation. And performing cross-frame Bayesian probability association by utilizing a Bayesian reasoning method to perform probability matching of texture features among time sequence images so as to remove feature points influenced by interference factors such as shielding, dynamic objects and the like. The Bayesian probability association is based on the prior distribution and the updating matching reliability of the observed data, and is a dynamic characteristic matching method with high robustness. Pose estimation data, which represents the position and pose (including position coordinates and orientation angles) of a vehicle at a certain moment, is used for navigation and path tracking. The module is oriented to multi-vehicle co-location application, and realizes the sharing and complementation of the positioning data among vehicles by introducing a communication topology modeling and confidence fusion strategy, and suppresses the accumulation and diffusion of the positioning errors of the single vehicle. Inertial navigation data is short-time position change information calculated based on inertial elements such as an accelerometer, a gyroscope and the like, but drift errors exist for a long time. Communication topology, which describes the information transmission structure, such as star, ring or mesh, between vehicles in a fleet, determines the path along which error information can travel. Error propagation chain model-modeling how error information propagates in a fleet of vehicles for estimating potential error propagation paths and superposition effects for each vehicle. And a multi-vehicle confidence weight fusion strategy, namely dynamically distributing confidence weights according to the stability of positioning data of each vehicle, the quality of a sensor and the stability of communication, and carrying out distributed filtering fusion so as to correct the differentiation errors among vehicles. Confidence weight, numerical index, used for reflecting the credibility of the data source. The anchor point space-time prediction module is used for predicting space-time position distribution of future reachable anchor points by combining with an operation plan in a port scheduling system, so that an anchor point activation strategy is optimized in advance, and prospective deployment of positioning service and dynamic scheduling of system resources are realized. Job planning analysis, namely reading data about future job tasks of equipment in a scheduling system, such as path planning, loading and unloading arrangement and the like. And modeling a motion state, namely constructing a future motion trail model based on parameters such as the current pose, the speed, the acceleration and the like of the vehicle. And the space-time reachability prediction model is used for predicting the possible passing area and the spatial range which can be covered by the anchor point in a certain future time period of the vehicle by integrating the operation plan and the motion model. And the anchor point activation priority is used for calculating the importance sequence of each candidate anchor point to the future positioning task according to the prediction model and indicating the anchor point scheduling priority. And (3) controlling instruction feedback, namely transmitting a priority decision result back to the dynamic anchor point module to form a task-driven closed-loop anchor point optimization system.

The method comprises the specific processes of analyzing the dynamic position data of the port mobile equipment in real time, obtaining real-time coordinates and moving directions of a portal crane boom and temporary traffic equipment, adjusting attenuation coefficients of a signal propagation model in a segmented mode according to the distribution density of metal obstacles in a container stacking area, generating a non-uniform attenuation gradient, calculating signal effective coverage boundaries from anchor point to anchor point according to the corrected attenuation gradient, generating a vector map described by polygon vertex coordinates, matching the real-time position of a vehicle with the vector map in a geometric relationship mode, marking the area with the signal strength lower than a positioning threshold as a coverage blind area, and outputting a blind area vertex coordinate list.

In the embodiment, the step of analyzing the dynamic position data of the port mobile equipment in real time illustrates that the dynamic position information of the port mobile equipment is obtained in real time through a vehicle-mounted positioning terminal (GNSS/IMU) and a positioning gateway deployed in a port scheduling system. The analysis data includes the following information of the vehicle number and type (AGV, tractor, empty van, etc.), the current coordinate point, the movement direction (unit vector formAnd heading angle) And the position and movement track of other dynamic facilities (such as gantry crane boom, guide vehicle and temporary obstacle). The steps of analyzing the distribution density of the metal barrier and adjusting the signal propagation model are described in that the space distribution characteristics of the metal objects such as a container stacking area, a crane structure and the like are obtained by utilizing a port three-dimensional modeling system or a laser radar point cloud reconstruction model. The attenuation degree of the signal propagation path is corrected according to the density of the metal objects and the shielding structure. Modeling of non-uniform attenuation, namely establishing a group of signal attenuation functions modified by sections according to the density distribution of metal barriers in a path. Setting: propagation power of the jth anchor point signal in the ith section path; The distance of the path segment; An environmental attenuation factor related to the metal shielding density of the segment. The segment signal power can be expressed as: And wherein: Initial transmitting power of j anchor point; And adjusting according to the on-site shielding condition, for example, setting the area of the high-density container to be 3.0-4.5, and setting the area of the empty container to be 2.0-2.5. The integral signal attenuation result from the anchor point j to any coordinate point M can be obtained through summation of a plurality of paths . The step of calculating the coverage boundary of the anchor point signal is described as that based on the two-dimensional geographical coordinate grid of the port, the ray scanning is carried out from each anchor point to each direction, and the signal intensity on the ray path is judged to be reduced to the threshold value according to the non-uniform modelIs the furthest point of the direction. And the polygon boundary construction mode is to connect boundary points in each direction to form an irregular polygon boundary representing the effective area of the anchor point signal. The polygon boundary is represented in the form of a list of vertex coordinates: Wherein, the method comprises the steps of, Is the firstThe signal coverage area of the individual anchor points,Is the number of boundary vertices. The step of constructing vector map and matching vehicle position is described as that based on anchor point polygon boundary, constructing real-time vector map of port and making current position of each vehicleProjected into a map. Utilizing point inpolygon algorithm (such as ray method and round number method) to judge that the vehicle is in effective coverage area of a certain anchor point or notThe vehicle is subjected to the firstThe anchor point signal is covered if the vehicle is not at randomIn, or at, the combined signal power of the pointAnd judging the point as an anchor point blind area. And marking blind areas and outputting blind area polygon coordinates, namely performing spatial clustering on all areas which are not covered by any anchor point effective signals, and extracting continuous blind area boundaries. Employing, for example, DBSCAN density clusteringOutputting a vertex coordinate list thereof: Wherein, the method comprises the steps of, Represent the firstPolygonal boundaries of the dead zone regions. The coordinate information is used as an input basis of a subsequent module (such as an anti-interference image sensing module) to guide the module to perform vision-aided positioning in the blind area.

In this embodiment, the steps of collecting and preprocessing vehicle history track data illustrate that a multi-day vehicle track log is retrieved from a port dispatch system, and the track data format is as follows: And wherein: First of all Historical track data of the vehicle; First of all The location of the individual time points; corresponding time stamps; the total number of track points recorded by the vehicle. And (5) carrying out time alignment and space normalization on all the vehicle track sets, and then sending the vehicle track sets into a clustering model. 2. The step of identifying the high-frequency passing area based on the density clustering algorithm is described as adopting a density-based spatial clustering algorithm (such as DBSCAN) to process the track point set so as to identify the high-frequency moving area of the vehicle. Let the track set be: and outputting the result after clustering: And wherein: First of all High density traffic clusters; the cluster will correspond to the specific harbor scene, such as the main channel of the yard and gate entrance and exit. The step of generating the anchor point activation priority instruction is to quantitatively analyze the clustering clusters according to the traffic frequency, the time period distribution and the traffic density, and calculate the priority score lambda_q of each cluster. The scoring model is as follows: And wherein: The traffic frequency of vehicles in the cluster; Average vehicle density in unit time; Time coverage rate of a traffic period in a cluster; Weight coefficient (set according to harbor district operation weight). According to From high to low ordering, generating an anchor priority activation inventoryGuiding the dynamic anchor point cooperative construction module to preferentially start the mobile anchor point corresponding to the high traffic area. The step of dynamically activating adjacent anchor points by combining the blind area information is described as follows: The current position of the vehicle Judging Euclidean distance with the dead zone boundary, and setting the nearest point of the dead zone boundary asThe following steps are: When (1) If the vehicle is about to enter the dead zone, the method sends out an early activation instruction to the anchor points of the mobile equipment near the boundary of the dead zone, and the vehicle is driven in at a speed according to the vehicle speedEstimating residence time in blind zoneSetting anchor delay closing time as current time plus. Dynamically updating an anchor activation schedule and coverage step description, establishing an anchor control scheduleAt time t, the activation state of each anchor point is represented: And wherein: S anchor point numbering; The predicted closing time is dynamically updated by combining the current position of the vehicle, the path prediction and the dead zone residence time, and the activation state 'on/off' is determined by whether the vehicle is at the dead zone boundary or not and the cluster high-priority zone. According to the change of the vehicle motion trail, the anchor point coverage priority and the working time of each anchor point are continuously updated, so that the blind area is ensured not to be missed by long-time coverage, and the energy consumption waste caused by redundant activation of anchor point resources is avoided.

The method comprises the specific processes of dividing a ground image acquired by a vehicle-mounted camera into a plurality of square local blocks in a coverage blind area range, calculating local gray entropy values of the blocks, screening blocks with gray entropy values higher than an environment background as candidate texture areas, comparing entropy value changes of the same blocks in adjacent frame images, eliminating abnormal high entropy blocks caused by instantaneous reflection or projection, performing morphological closing operation on the candidate texture areas, filling the broken areas, and generating continuous positioning reference masks.

In the embodiment, the steps of dividing the image area and calculating the local gray entropy illustrate that the gray image acquired by the vehicle-mounted camera is processedDivided into a plurality of non-overlapping partial square blocks: And wherein: the ground gray level image of the current frame; First of all Line 1Image local blocks of columns; the dimension of the divided block grid is determined according to the image size and the block granularity. Local gray entropy calculation for each block Calculating the gray entropy value:And wherein: Gray scale (typically 256); Gray value is The pixels of (a) are in a blockIs a normalized frequency of (a); To prevent logarithmic terms from appearing Is a small constant (e.g). The step of screening candidate texture blocks and eliminating high entropy abnormality is described as 1. Background entropy threshold setting, calculating entropy average value of all blocks of the whole imageAnd standard deviationSetting a background entropy threshold: And wherein: adjusting the coefficient, controlling the screening intensity (generally taking 1-2), and regarding the area exceeding the threshold as a candidate block with texture distinction. 2. Inter-frame abnormal high entropy rejection for continuous two-frame images And (3) withComparing entropy differences of the same-position blocks: If (1) Then the change is judged to be unstructured disturbance (such as reflection, projection) and the block is removed.Setting a threshold value, reflecting the maximum entropy change (experience value is 0.5-1.0) of stable textures, and retaining stable and repeatable texture blocks after eliminating. 3. The morphological closing operation repair and mask generation steps are described as binary mask map of candidate texture blockMorphological closing operations (first expansion then corrosion) were performed to eliminate small holes, junction break areas: Dilate, erode, which respectively represent image swelling and etching operations; Structural elements (e.g Square kernel) and the operation result is to close continuous positioning reference mask images. Outputting final result of continuous texture positioning maskAnd the mask is used as a texture region suitable for positioning and matching in the frame image, and is sent to a subsequent cross-frame Bayesian attitude estimation module after being aligned with the coverage information of the anchor point signal in time and space. The vehicle can still rely on stable ground texture to carry out auxiliary positioning in the blind area, and positioning robustness is improved.

The method comprises the specific processes of carrying out space-time calibration on stable texture features in continuous multi-frame images, recording feature point motion tracks, constructing a cross-frame feature association probability matrix, calculating the association probability of current frame features and historical frames according to a Bayesian inference model, eliminating the features if the probability is lower than a dynamic interference threshold value, aligning the screened features with anchor signal time stamps, solving vehicle pose through a perspective transformation matrix, fusing multi-frame pose estimation results, and generating smoothed vehicle track data.

In the embodiment, the steps of space-time calibration and feature track construction of the stable texture features are described as extracting a key feature point set of the stable texture region from a mask map with gray entropy segmentation and morphological processing completed: And wherein: Image frame time index; First of all Image coordinates of the feature points; Time of the event And a texture feature point set extracted at the moment. Tracking the same points in continuous multiframes by using an optical flow method or a feature matching algorithm to construct a motion track sequence of each feature point: . The step of modeling the cross-frame feature association probability and the step of Bayesian inference show that for each pair of feature point tracks, the cross-frame feature association probability is constructed based on factors such as the similarity of the image space positions, the consistency of the motion directions, the time interval attenuation factors and the like: And wherein: First of all Is characterized in thatTime and the firstIs characterized in thatBayesian association probability of time; Likelihood function, measure position and motion difference; The prior probability is set according to the regional texture stability, the denominator is a normalization factor, and the sum of the association probabilities of all candidate feature points is ensured to be 1. Note that likelihood functions may be defined as two-dimensional gaussian distributions, and specific formulas are omitted and can be found in the literature in the field of image matching. If the maximum association probability of a certain characteristic point is lower than the dynamic interference threshold value Then the point is determined to be the source of dynamic interference (e.g., pedestrian, robotic arm) and culled: . The time synchronization of the time alignment with the anchor point signal and the vehicle pose calculation step is that the retained stable characteristic points are cut off at the time of image acquisition Time-cut with anchor point signalAnd (3) aligning, and performing time fusion by adopting a linear interpolation or nearest neighbor matching mode. Solving a single-frame perspective transformation matrix by utilizing the screened plane characteristic points and the corresponding physical coordinates thereof on the ground of a real port and passing through a homography matrixEstablishing a mapping relation between an image and world coordinates: And wherein: characteristic points in an image coordinate system are aligned; corresponding to the position under the world coordinates; perspective transformation matrix based on matching point pair estimation by Two-dimensional pose of thrust reverser body. Multiple frames of pose fusion generating track step description, namely estimating vehicle pose for each frameAnd (3) performing track smoothing by adopting a weighted filtering strategy in a sliding window: And wherein: Smoothing the weight, and satisfying the condition that the weight of the center frame is higher (such as Gaussian weight or triangle kernel); sliding window radius; Smoothing the filtered pose, and continuing The component tracks are used for subsequent error suppression module processing.

Specifically, the construction logic for receiving pose estimation data and vehicle inertial navigation data and constructing an error propagation chain model by combining the communication topological relation among vehicles is as follows. According to the consistency of vehicle motion direction, the transmission attenuation coefficient of the error along the communication link is calculated to generate an error propagation chain matrix, and when the vehicle position or the communication topology changes, the error propagation weight and the attenuation coefficient are recalculated.

In the present embodiment, communication topology construction and connection state maintenance are performed, and a dynamic communication topology construction is performed based on a wireless communication signal strength (RSSI) and a maximum communication distance threshold between vehicles in a fleetDynamically establishing mesh communication topologyWherein: representing all vehicle nodes in a motorcade; Representing a set of vehicle pairs meeting communication conditions, communication edges If and only if: And wherein: Vehicle(s) And (3) withIs the euclidean distance of (2); A lower threshold of the communication signal strength. The node connection state table updates the heartbeat packet of each vehicle node timing broadcast state, and the central controller or the edge node gathers the communication state to generate the topology connection state table Wherein: . The bicycle error node represents and adjacent error weight is calculated, and the bicycle error node defines the current inertial navigation accumulated error of each bicycle Represented as node error values in the graph, wherein: Representing a position residual error between the inertial solution and the visual pose; Vehicle(s) The current pose is calculated through inertial navigation; Vehicle(s) The computed pose is aligned by vision/anchor points. The adjacent error propagation weight aims at any pair of adjacent vehicles) Defining error propagation weightsIndicating error slave vehiclesTransferred to vehicleReliability of (2): And wherein: Error amplification factor (set according to system experience); The method is characterized in that the Euclidean distance between visual estimation pose is represented, the larger the weight value is, the closer the pose of two vehicle is represented, and the error can be effectively propagated and compensated. Modeling of motion direction consistency and error transfer attenuation, and setting of vehicle by direction consistency factor 、The motion direction vectors of (a) are respectively: and then the two direction consistency factors are defined as: If the directions of the two vehicles are consistent (collinear and the same direction), If the directions of the two directions are opposite,. The error attenuation coefficient is calculated by comprehensively considering the direction consistency and the communication intensity, and the error attenuation coefficient is defined as follows: If the directions of the vehicles are inconsistent ) The attenuation coefficient is 0, and the attenuation coefficient of the effective link determines the efficiency of error transmission inside the motorcade. Error propagation chain matrix construction and dynamic update mechanism, the error propagation chain matrix defines an error propagation chain matrix: the elements are as follows: each row represents a vehicle Error may be propagated to the attenuation intensity of other vehicles, and error tracing, weighted feedback and fusion can be realized on the matrix diagram. Dynamic recalculation mechanism when the position of any one vehicle changes beyond a thresholdOr the communication topology changes (newly added/disconnected edges), re-executing the two steps of reconstructing the communication diagramUpdating the connection state table,,Reconstruction of。

The method comprises the specific processes of correcting accumulated positioning errors through a multi-vehicle confidence weight fusion strategy, namely dynamically distributing confidence weights according to the alignment degree and the historical positioning stability of vehicle pose estimation data and anchor point signals, reducing abnormal vehicle weights and triggering local texture matching if the adjacent vehicle pose difference exceeds a set threshold value, fusing multi-vehicle pose data through a weighted average algorithm based on an error propagation chain matrix to restrain single vehicle errors, feeding back fused poses to each vehicle node, and updating an inertial navigation initial state.

In this embodiment, the confidence weight assignment logic, dynamic confidence index calculation, is for any vehicle node in the fleetLet its current visual pose estimate beInertial navigation pose isThe matching degree of the anchor point signals is as followsHistorical stability score ofThen the vehicle confidence weightDefined as follows: And wherein: Representing a vehicle The alignment degree of the current pose and the anchor point signal; representing the positioning stability (such as inverse ratio of the variance of the position fluctuation) of the vehicle in a short period; Representing a vehicle Neighbor vehicle set under communication topology, denominator normalization ensures. Note that the method embodies a weight evaluation mechanism based on anchor point matching quality and historical performance, and avoids pollution to integral positioning caused by a single abnormal node. Abnormality detection and local texture compensation triggering, pose difference detection for any adjacent vehicle、The current visual pose difference value is calculated as follows: And if present: And then identify Or (b)In the presence of an anomaly, triggering operations of temporarily reducing the confidence level of the anomalous vehicle node (e.g., multiplied by an adjustment factor) And initiating a local texture repositioning process of the abnormal node, performing texture matching verification on the latest frames of images, and recovering the confidence level if repositioning is effective, otherwise, continuously reducing the influence weight of the node in the fusion model. Weighting fusion correction based on propagation chain matrix, and multi-vehicle pose fusion model for target vehicleThe neighborhood vehicles are gathered intoError propagation chain matrix isFused pose estimationThe calculation is as follows: And wherein: attenuation coefficients in the error propagation chain (see section above for details); confidence based on anchor point matching and historical stability assessment; Adjacent vehicle The drifting error or local failure of a certain vehicle can be obviously restrained after the weighted fusion of multiple vehicles. The equalization normalization strategy is to avoid weight sum unbalance, and perform normalization processing before fusion: . The pose feedback and inertial navigation state reinitialization are used for merging the smooth pose As a high confidence reference result, fed back to the vehicleThe local navigation system is used for updating the initial state of inertial navigation, and the specific updating formula is as follows: And wherein: Initializing updated inertial navigation positions; Inertial navigation speed vector estimation; time interval of two navigation status updates.

The method comprises the steps of analyzing an operation plan of mobile equipment in a port scheduling system, combining the motion state of a corrected vehicle pose to construct a space-time reachability prediction model, extracting operation time table and path planning data of the mobile equipment in the port scheduling system, including a portal crane boom moving track and a temporary equipment deployment period, constructing a vehicle kinematics model according to the corrected pose data, predicting a space-time region which can be reached by the vehicle in a future period, carrying out intersection operation on the space-time distribution of the vehicle prediction track and a mobile equipment anchor point to generate a space-time reachability probability distribution map, generating an anchor point activation priority list according to the space-time density of the intersection region, and marking the high-probability touch anchor point.

In this embodiment, the operation plan and the path planning data are analyzed, the operation schedule extracts an operation schedule of a mobile device (such as a gantry crane boom and a temporary traffic device) from a port scheduling system, and the time interval set is defined as: Wherein Indicating the job start and end time nodes. The path planning data acquisition comprises a set of moving track points of the equipment: Wherein Representing position coordinates in three-dimensional space. Vehicle kinematics model construction 1. Input corrected pose data to vehicleTo correct the pose dataAs a starting point, a state vector is established: And wherein: two-dimensional position coordinates of the vehicle; the course angle of the vehicle; The speed of the vehicle. Kinematic prediction formula for predicting future time by using vehicle kinematic model Position: And wherein: Vehicle angular velocity (estimated from historical data); predicting a time step. Generating a space-time reachability probability distribution map, and constructing a predicted track through a kinematic model to generate a predicted track set of a future period [ T, t+T ] of the vehicle: the space-time anchor distributing mobile device anchor space-time position set is as follows: Wherein As the spatial coordinates of the anchor point,The time interval is activated for the anchor point. Space-time intersection calculation is carried out on the vehicle prediction track and anchor point space-time distribution, and the vehicle is calculated at each momentTouch anchor pointProbability of (2)Combining spatial distance with time overlap relationship: And wherein: Representing euclidean distance; The tolerance radius of the space position reflects the allowable error range; As a function of the readiness And 1 if not, and 0 if not. Integrating all anchor points and prediction time periods by using the space-time reachability probability distribution map to generate a probability matrix: . Anchor point activation priority list generation, space-time density calculation according to space-time probability matrix Calculating the total probability of reaching the anchor point in the prediction period: Priority ordering From big to small ordering, an anchor activation priority list is generated: . High probability anchor marking is for meeting a threshold condition Is marked as a high-priority activated anchor point and is used as a control input of the dynamic anchor point cooperative construction module.

The method comprises the specific processes of predicting a time-space distribution matrix of future reachable anchors, generating a control instruction according to the activation priority of the anchors in the matrix and feeding back the control instruction to a dynamic anchor cooperative construction module, wherein the time-space reachability probability distribution map is quantized into the matrix, the dimension of the matrix is a time-space-anchor identifier, the order of activating the priority of the anchors is generated according to the probability value sequence of space-time nodes in the matrix, the high-probability reachable anchors are preferentially awakened, the priority instruction is fed back to the dynamic anchor cooperative construction module through a port wireless private network, anchor awakening or dormancy operation is triggered, anchor activating effects are monitored in real time, and if a vehicle does not reach the anchors according to the prediction, the priority instruction is dynamically adjusted and the matrix is rebuilt.

In this embodiment, the space-time reachability probability distribution map is quantized into a matrix, and the probability distribution of each anchor point that the port vehicle may reach in the space-time range in the future is converted into a three-dimensional matrix representation, and the dimensions of the matrix include a time sequence, a space region and an anchor point identifier. Let the matrix beWherein: representing a discrete set of time nodes, in an amount of ;Representing a set of spatially partitioned regions, the number being;Representing anchor point set, the number isMatrix elementsIs shown at the time pointSpatial regionInner vehicle touch anchor pointIs a probability of (2).Calculating anchor activation priority based on matrixComprehensive touch probability of each anchor point in the plurality of anchor points, and calculating the anchor pointIs (are) activation priority index:And wherein: representing space-time nodes Reflecting the priority or importance of the spatio-temporal node (e.g., node weight is greater near the current time and the weight of the key operation area is greater); Is an anchor point Is included. Anchor activation priority list based onDescending order of priority activationHigher anchor points. Generating and issuing anchor point control instructions according to the activation priority thresholdFor each anchor pointGenerating an activation control instruction: And wherein: represents the active state of the anchor point, 1 is active, 0 is dormant, threshold value Can be dynamically adjusted according to the system load and the resource condition. And sending the control instruction to the port wireless private network, feeding back the control instruction to the dynamic anchor point cooperative construction module, and executing corresponding anchor point awakening or dormancy operation. The real-time monitoring and dynamic adjusting system continuously monitors the anchor point activating effect, records the actual condition of the vehicle touching the anchor point, and if the vehicle does not touch the anchor point according to the prediction, adjusts the activating priority weight or threshold according to the actual data to regenerate the space-time distribution matrixAnd updates the activation instruction. This process can be represented by the following correction mechanism: And wherein: predicting probability for the previous period; Probability estimation updated based on actual monitoring data; historical predictions are weighed against real-time adjustments for smoothing coefficients.

Referring to FIG. 2, the method for positioning internal and external vehicle fleets in a port comprises the following steps of S1, acquiring dynamic position data of port mobile equipment in real time, dynamically calculating anchor point signal coverage according to signal propagation attenuation characteristics in a metal environment, generating anchor point activating instructions and anchor point coverage blind area information by combining vehicle history track clustering analysis, S2, receiving the anchor point coverage blind area information, carrying out local gray entropy segmentation on ground images acquired by a vehicle-mounted camera, screening high-resolution texture areas, eliminating dynamic interference through cross-frame Bayesian probability association, generating pose estimation data aligned with anchor point signal time and time, S3, constructing an error propagation chain model by combining with inter-vehicle communication topological relation, correcting accumulated positioning errors through a multi-vehicle confidence weight fusion strategy, S4, analyzing an operation plan of mobile equipment in a port scheduling system, constructing a time-space accessibility prediction model by combining with the motion state of corrected vehicle pose, predicting a time-space distribution matrix of future reachable anchor points, generating control instructions according to the anchor point activating priority, and feeding back to the step S1.

In the embodiment, step S1, the real-time adjustment of the dynamic anchor point signal coverage breaks through the traditional fixed anchor point coverage assumption, aims at the dynamic modeling of the signal attenuation characteristics in the metal dense environment, accurately describes the non-uniform propagation of the signal, realizes the real-time dynamic calculation of the coverage, and remarkably improves the space effectiveness of the positioning signal. The clustering analysis of the vehicle history tracks is integrated, the density clustering is used for creatively identifying the high-frequency passing area, the space-time priority scheduling of anchor point activation is realized, the resource allocation is effectively optimized, invalid awakening is avoided, and the energy efficiency and response speed of the system are improved. The blind area information feedback mechanism realizes the accurate positioning of the coverage blind area, dynamically activates adjacent anchor points through the vertex coordinates of the blind area, compensates signal blind areas, and enhances the continuity and stability of a positioning system. And S2, the local gray entropy segmentation positioning reference area adopts a local gray entropy method to screen high-distinction textures, so that the stability and the recognition rate of visual features are improved, and illumination and texture changes in a complex harbor environment are effectively treated. Introducing a Bayesian inference mechanism into the cross-frame Bayesian probability association model, performing space-time consistency verification on texture features in continuous frames, innovatively removing dynamic interference factors such as reflection and projection, and guaranteeing the accuracy and the robustness of pose estimation data. And a fusion mechanism of space-time alignment synchronizes visual features with the anchor point signal time stamp, fuses multi-mode information and improves the overall positioning accuracy. And step S3, constructing a dynamic communication topology based on the wireless communication signal intensity to reflect the network connection state of the vehicle in real time, dynamically establishing an error propagation chain of multiple workshops, breaking through the positioning limitation of a bicycle and realizing error propagation and control across the vehicle. The error propagation chain model innovation maps the inertial navigation accumulated errors into directed graph nodes, calculates error propagation weights and attenuation coefficients by using communication topology and motion consistency, and scientifically reveals error propagation paths and influence degrees. The vehicle positioning confidence coefficient is dynamically distributed by the multi-vehicle confidence coefficient weight fusion strategy, abnormal vehicle errors are restrained through weighted fusion, the stability and reliability of overall vehicle team positioning are improved, and collaborative error correction is achieved. And S4, the scheduling operation plan and the depth fusion of the motion state are used for incorporating the port scheduling system operation plan into a positioning prediction flow for the first time, and a vehicle motion model for correcting the pose is combined for accurately predicting a space-time area reached by a future anchor point, so that prospective positioning resource scheduling is realized. The space-time reachability prediction model quantifies the contact probability of the anchor points based on intersection operation of track prediction and anchor point space-time distribution, systematically builds a space-time distribution matrix and scientifically guides the activation priority ordering of the anchor points. The closed-loop control feedback mechanism adjusts the dynamic anchor point cooperative construction module in real time through the generated control instruction, so that the self-adaptive adjustment of the positioning system is realized, and the real-time response capability and accuracy of the positioning system are improved.

In summary, the present application has at least the following effects:

The system and the method for positioning the internal and external motorcades of the port effectively inhibit the influence of signal propagation attenuation and dynamic interference in a metal environment by combining the dynamic anchor point cooperative construction with the anti-interference environment feature extraction, and realize more accurate vehicle pose estimation. Based on the cross-frame Bayesian probability association, the dynamic interference features are removed, an error propagation chain model is built by combining the inter-vehicle communication topology, and the multi-vehicle confidence weight fusion strategy effectively suppresses the accumulated errors of the inertial navigation of the single vehicle, so that the stability and the robustness of the overall positioning system are improved. Dynamic adjustment of anchor point activation is achieved through vehicle history track clustering and space-time reachability prediction, invalid dissipation of anchor point resources is avoided, signal coverage is optimized, and system energy consumption and operation and maintenance cost are reduced. Based on a closed loop feedback mechanism of the space-time distribution matrix and the dynamic priority order, the real-time adjustment and optimization of the anchor point activation strategy are realized, and the quick response capability of the system to the port operation environment change is ensured. And a mesh communication topology and error propagation chain model is adopted, the multi-vehicle pose data are fused, the multi-node co-location requirement of the large-scale motorcade of the port is met, and the overall operation efficiency and the safety management level of the motorcade are improved.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of systems, apparatuses (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

1. The system for positioning the motorcades inside and outside the port is characterized by comprising a dynamic anchor point cooperative construction module, an anti-interference environment characteristic extraction module, a motorcade cooperative error suppression module and an anchor point space-time prediction module;

The dynamic anchor point cooperative construction module is used for acquiring dynamic position data of the port mobile equipment in real time, dynamically calculating an anchor point signal coverage range according to signal propagation attenuation characteristics in a metal environment, and generating an anchor point activating instruction and anchor point coverage blind area information by combining vehicle historical track cluster analysis;

The anti-interference environment characteristic extraction module is used for receiving anchor point coverage blind area information, carrying out local gray entropy segmentation on a ground image acquired by the vehicle-mounted camera, screening a high-resolution texture area, eliminating dynamic interference through cross-frame Bayesian probability association, and generating pose estimation data aligned with anchor point signal time and space;

The motorcade cooperative error suppression module is used for receiving pose estimation data and vehicle inertial navigation data, constructing an error propagation chain model by combining communication topological relation among vehicles, and correcting accumulated positioning errors through a multi-vehicle confidence weight fusion strategy;

The anchor point space-time prediction module is used for analyzing an operation plan of mobile equipment in the port scheduling system, constructing a space-time reachability prediction model by combining the motion state of the corrected vehicle pose, predicting a space-time distribution matrix of future reachable anchor points, generating a control instruction according to the anchor point activation priority in the matrix, and feeding back the control instruction to the dynamic anchor point cooperative construction module.

2. The system for locating internal and external fleets in port according to claim 1, wherein the method comprises the following steps of:

analyzing dynamic position data of port mobile equipment in real time, and acquiring real-time coordinates and movement directions of a portal crane boom and temporary traffic equipment;

According to the distribution density of metal barriers in a container stacking area, the attenuation coefficient of a signal propagation model is adjusted in a segmented mode, a non-uniform attenuation gradient is generated, according to the corrected attenuation gradient, the effective coverage boundary of a signal is calculated by anchor points, a vector map described by polygon vertex coordinates is generated, the real-time position of a vehicle is matched with the vector map in geometric relation, the area with the signal strength lower than a positioning threshold is marked as a coverage blind area, and a blind area vertex coordinate list is output.

3. The system for locating internal and external fleets in port according to claim 2, wherein the specific process of generating anchor point activating instruction and anchor point coverage blind area information by combining vehicle history track cluster analysis is as follows:

collecting historical track data of a vehicle, and identifying a high-frequency passing area comprising a container transfer area and a gate channel through a density clustering algorithm;

Generating an anchor point activation priority order according to the clustering result, and preferentially activating the anchor points of the mobile equipment in the high-frequency area;

And matching the coordinates of the coverage blind area with the real-time position of the vehicle, activating the anchor points of adjacent mobile equipment according to the vertex coordinates of the blind area when the vehicle drives into the boundary of the blind area, prolonging the working time of the anchor points, updating the anchor point activation time sequence table according to the real-time position of the vehicle, and adjusting the signal coverage.

4. The port internal and external fleet positioning system according to claim 3, characterized in that the specific process of local gray entropy segmentation of the ground image collected by the vehicle-mounted camera and screening the high-resolution texture area is as follows:

Dividing a ground image acquired by a vehicle-mounted camera into a plurality of square local blocks in a coverage blind area range, calculating local gray entropy values of the blocks, screening blocks with gray entropy values higher than an environment background as candidate texture areas, comparing entropy value changes of the same blocks in adjacent frame images, and eliminating abnormal high entropy blocks caused by instantaneous reflection or projection;

And performing morphological closing operation on the candidate texture region, filling the fracture region, and generating a continuous positioning reference mask.

5. The system for locating internal and external fleets in port according to claim 4, wherein dynamic interference is eliminated through cross-frame Bayesian probability association, and the specific process of generating pose estimation data aligned with anchor point signal time and space is as follows:

Performing space-time calibration on stable texture features in continuous multi-frame images, recording motion tracks of feature points, constructing a cross-frame feature association probability matrix, calculating association probability of current frame features and historical frames according to a Bayesian inference model, and eliminating the features if the probability is lower than a dynamic interference threshold;

And aligning the filtered characteristics with the anchor point signal time stamp, calculating the vehicle pose through a perspective transformation matrix, and fusing multi-frame pose estimation results to generate smoothed vehicle track data.

6. The system for locating a fleet of vehicles inside and outside a port according to claim 5, wherein the construction logic for constructing an error propagation chain model by combining the pose estimation data and the vehicle inertial navigation data and the communication topological relation among vehicles is as follows.

According to the wireless communication signal intensity and the distance threshold value between vehicles, dynamically establishing a mesh communication topology, maintaining a vehicle node connection state table in real time, mapping the accumulated errors of the inertial navigation of the single vehicle into directed graph nodes, and constructing error propagation weights through the pose differences of adjacent vehicles;

According to the consistency of the vehicle movement direction, the transmission attenuation coefficient of the error along the communication link is calculated, an error propagation link matrix is generated, and when the vehicle position or the communication topology changes, the error propagation weight and the attenuation coefficient are recalculated.

7. The port internal and external fleet positioning system according to claim 6, characterized in that the specific process of correcting the accumulated positioning error by the multi-vehicle confidence weight fusion strategy is as follows:

dynamically distributing confidence weight according to the alignment degree of the vehicle pose estimation data and the anchor point signal and the historical positioning stability;

If the pose difference of the adjacent vehicles exceeds a set threshold, reducing the weight of the abnormal vehicles and triggering local texture matching, and based on an error propagation chain matrix, fusing multi-pose data through a weighted average algorithm to inhibit single vehicle errors;

And feeding the fused pose back to each vehicle node, and updating the initial state of inertial navigation.

8. The system for locating internal and external fleets in port according to claim 7, wherein the construction logic for analyzing the operation plan of the mobile device in the port dispatching system and constructing the space-time reachability prediction model by combining the motion state of the corrected vehicle pose is as follows:

extracting operation time schedule and path planning data of mobile equipment in a port scheduling system, wherein the operation time schedule and path planning data comprise a gantry crane boom moving track and a temporary equipment deployment period;

according to the corrected pose data, a vehicle kinematic model is constructed, a space-time region reachable by the vehicle in a future period is predicted, intersection operation is carried out on a vehicle predicted track and space-time distribution of anchor points of mobile equipment, and a space-time reachability probability distribution map is generated;

and generating an anchor point activation priority list according to the space-time density of the intersection region, and marking the high-probability touch anchor points.

9. The system for locating internal and external fleets in a port according to claim 8, wherein the specific process of predicting a space-time distribution matrix of future reachable anchor points, generating a control instruction according to the activation priority of the anchor points in the matrix and feeding back the control instruction to the dynamic anchor point collaborative building module is as follows:

Quantifying the space-time reachability probability distribution map into a matrix, wherein the dimension of the matrix is a time-space-anchor point identifier, and generating an anchor point activation priority instruction according to probability value ordering of space-time nodes in the matrix, so as to wake up the high-probability touch anchor point preferentially;

And feeding the priority order back to the dynamic anchor point cooperative construction module through the port wireless private network, triggering anchor point awakening or dormancy operation, monitoring anchor point activating effect in real time, and dynamically adjusting the priority order and regenerating the matrix if the vehicle does not reach the anchor point according to prediction.

10. The method for locating the internal and external fleets of the port, which is applied to the internal and external fleets of the port locating system of any one of claims 1 to 9, is characterized by comprising the following steps:

s1, acquiring dynamic position data of port mobile equipment in real time, dynamically calculating an anchor point signal coverage area according to signal propagation attenuation characteristics in a metal environment, and generating an anchor point activating instruction and anchor point coverage blind area information by combining vehicle history track cluster analysis;

S2, receiving anchor point coverage blind area information, carrying out local gray entropy segmentation on a ground image acquired by a vehicle-mounted camera, screening a high-discrimination texture area, and eliminating dynamic interference through cross-frame Bayesian probability association to generate pose estimation data aligned with anchor point signal time and space;

S3, pose estimation data and vehicle inertial navigation data are combined with communication topological relations among vehicles to construct an error propagation chain model, and accumulated positioning errors are corrected through a multi-vehicle confidence weight fusion strategy;

S4, analyzing an operation plan of the mobile equipment in the port scheduling system, constructing a space-time reachability prediction model by combining the motion state of the corrected vehicle pose, predicting a space-time distribution matrix of future reachable anchor points, generating a control instruction according to the anchor point activation priority in the matrix, and feeding back to the step S1.