Port internal and external vehicle fleet positioning system and method

Abstract

The present invention discloses a system and method for positioning fleets inside and outside ports, and relates to the field of intelligent positioning technology. It is used to solve the problems of frequent positioning blind spots, sensitivity to dynamic interference, uncontrollable error accumulation, and low efficiency in anchor resource scheduling in complex port scenarios. The anchor point coverage is dynamically calculated through the signal attenuation characteristics of the metal environment, and the blind spot information is generated in combination with vehicle trajectory clustering to optimize signal resource allocation. Ground texture grayscale entropy segmentation and cross-frame Bayesian correlation are performed in the blind spot to eliminate interference and generate synchronized posture data, thereby improving the robustness of visual positioning. An error propagation chain model is constructed based on the inter-vehicle communication topology, and the confidence weights of multiple vehicles are integrated to correct the accumulated error and enhance the consistency of fleet positioning. Finally, the port scheduling plan and vehicle motion status are analyzed, and a spatiotemporal prediction model is constructed to dynamically generate anchor point instructions to realize on-demand scheduling of anchor point resources. Significantly improve the positioning accuracy and resource utilization efficiency of port vehicles, and provide reliable support for port automation operations.

Description

Technical Field

[0001] The present invention relates to the field of intelligent positioning technology, and in particular to a system and method for positioning fleets inside and outside a port. Background Art

[0002] With the continuous improvement of the automation and intelligence level of global ports, container terminals are gradually transforming to an operating model of "paperless scheduling, unmanned handling, and information-based collaboration." The container trucks, forklifts, and automated guided vehicles (AGVs) operating vehicles within the port, as well as the logistics trucks outside the port, form a high-frequency collaborative operation network, which places higher demands on the real-time positioning accuracy of vehicles, system robustness, and scheduling response capabilities. In a typical port operating environment, due to the dense metal equipment and severe signal obstruction, the traditional positioning method that relies solely on satellite navigation has poor stability and large error fluctuations, making it difficult to support complex scheduling tasks and automated operating processes. In order to ensure operational safety and improve throughput efficiency, it is urgent to establish a high-precision multi-vehicle collaborative positioning system with dynamic adaptive capabilities for port scenarios.

[0003] Existing port vehicle positioning methods primarily fall into three categories: absolute positioning systems based on the Global Navigation Satellite System (GNSS); auxiliary positioning systems that rely on ground-based ultra-wideband (UWB) signal positioning, radio frequency identification (RFID) tag recognition, or anchor markers such as magnetic pins; and feature-based positioning systems based on vision and lidar. However, these methods suffer from the following substantial technical deficiencies in practical applications: First, anchor-based positioning solutions often employ fixed deployments and lack a mechanism for real-time coordination with vehicle operating paths, resulting in low anchor signal utilization and the formation of signal blind spots in dynamic operating areas. Second, visual positioning methods generally employ single-frame image feature extraction, which is susceptible to interference factors such as dynamic occlusions and lighting changes, and cannot stably output continuous pose estimation results. Third, multi-vehicle collaborative positioning systems generally employ a centralized processing architecture that fails to consider differences in communication relationships and positioning uncertainty between nodes within a fleet, leading to errors that easily accumulate and propagate within the system. Fourth, the anchor point activation strategy is mostly triggered by a fixed-period polling method, which is not linked to the operation plan in the port scheduling system in real time. It is impossible to realize the anchor point coverage prediction based on the future movement trend of the vehicle. The scheduling control response is delayed, affecting the overall collaborative performance of the system. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a system and method for positioning a fleet inside and outside a port, which solves the problems of the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a port internal and external fleet positioning system, including the following modules: a dynamic anchor point collaborative construction module, an anti-interference environment feature extraction module, a fleet collaborative error suppression module, and an anchor point spatiotemporal prediction module; the dynamic anchor point collaborative construction module is used to obtain the dynamic position data of the port mobile equipment in real time, dynamically calculate the anchor point signal coverage range according to the signal propagation attenuation characteristics under the metal environment, and generate the anchor point activation instruction and anchor point coverage blind area information in combination with the vehicle historical trajectory cluster analysis; the anti-interference environment feature extraction module is used to receive the anchor point coverage blind area information, perform local grayscale entropy segmentation on the ground image collected by the vehicle-mounted camera, and screen High-resolution texture areas are identified, and dynamic interference is eliminated through cross-frame Bayesian probability association to generate pose estimation data that is spatiotemporally aligned with the anchor point signal; the fleet collaborative error suppression module is used to receive pose estimation data and vehicle inertial navigation data, and construct an error propagation chain model based on the communication topology relationship between vehicles, and correct the accumulated positioning error through the multi-vehicle confidence weight fusion strategy; the anchor point spatiotemporal prediction module is used to parse the operation plan of mobile equipment in the port scheduling system, and construct a spatiotemporal reachability prediction model based on the motion state of the corrected vehicle pose, predict the spatiotemporal distribution matrix of future reachable anchor points, and generate control instructions based on the anchor point activation priority in the matrix and feed them back to the dynamic anchor point collaborative construction module.

[0006] Furthermore, the dynamic position data of the mobile equipment in the port is obtained in real time, and the specific process of dynamically calculating the anchor point signal coverage range according to the signal propagation attenuation characteristics in the metal environment is as follows: the dynamic position data of the mobile equipment in the port is analyzed in real time to obtain the real-time coordinates and movement direction of the gantry crane arm and temporary traffic equipment; according to the distribution density of metal obstacles in the container stacking area, the attenuation coefficient of the signal propagation model is adjusted in sections to generate a non-uniform attenuation gradient; based on the corrected attenuation gradient, the effective coverage boundary of the signal is calculated anchor by anchor point, and a vector map described by polygon vertex coordinates is generated; the real-time position of the vehicle is geometrically matched with the vector map, and the area with signal strength lower than the positioning threshold is marked as a coverage blind spot, and a list of blind spot vertex coordinates is output.

[0007] Furthermore, the specific process of generating anchor point activation instructions and anchor point coverage blind spot information based on vehicle historical trajectory clustering analysis is as follows: collect vehicle historical trajectory data, and identify high-frequency traffic areas, including container transfer areas and gate channels, through density clustering algorithm; generate anchor point activation priority instructions based on clustering results, and give priority to activating mobile device anchor points in high-frequency areas; match the coverage blind spot coordinates with the vehicle's real-time position. When the vehicle enters the blind spot boundary, activate the adjacent mobile device anchor points according to the blind spot vertex coordinates, and extend their working time. Update the anchor point activation timing table according to the vehicle's real-time position to adjust the signal coverage range.

[0008] Furthermore, the ground image captured by the vehicle-mounted camera is segmented by local grayscale entropy, and the specific process of screening high-resolution texture areas is as follows: within the coverage blind area, the ground image captured by the vehicle-mounted camera is divided into multiple square local blocks, the local grayscale entropy value of each block is calculated, and the blocks with grayscale entropy values higher than the environmental background are screened as candidate texture areas. The entropy value changes of the same blocks in adjacent frame images are compared, and abnormal high-entropy blocks caused by instantaneous reflection or projection are eliminated; morphological closing operations are performed on the candidate texture areas, the broken areas are filled, and a continuous positioning reference mask is generated.

[0009] Furthermore, dynamic interference is eliminated through cross-frame Bayesian probability association to generate pose estimation data that is temporally and spatially aligned with the anchor signal. The specific process is as follows: stable texture features in consecutive multi-frame images are temporally and spatially calibrated, the motion trajectory of feature points is recorded, a cross-frame feature association probability matrix is constructed, and the association probability of the current frame feature and the historical frame is calculated based on the Bayesian inference model. If the probability is lower than the dynamic interference threshold, the feature is eliminated; the filtered features are aligned with the anchor signal timestamp, the vehicle pose is solved through the perspective transformation matrix, and the multi-frame pose estimation results are fused to generate smoothed vehicle trajectory data.

[0010] Furthermore, the model receives pose estimation data and vehicle inertial navigation data, and constructs an error propagation chain model based on the inter-vehicle communication topology. The model's logic is as follows: Based on inter-vehicle wireless communication signal strength and distance thresholds, a mesh communication topology is dynamically established. A table of vehicle node connections is maintained in real time. The accumulated inertial navigation errors of individual vehicles are mapped to directed graph nodes, and error propagation weights are constructed based on the pose differences between adjacent vehicles. Based on the consistency of vehicle motion directions, the error transmission attenuation coefficient along the communication link is calculated to generate an error propagation chain matrix. The error propagation weights and attenuation coefficients are recalculated when the vehicle position or communication topology changes.

[0011] Furthermore, the specific process of correcting the accumulated positioning error through the multi-vehicle confidence weight fusion strategy is as follows: confidence weights are dynamically assigned based on the alignment between the vehicle pose estimation data and the anchor point signal, as well as the historical positioning stability; if the pose difference between adjacent vehicles exceeds the set threshold, the weight of the abnormal vehicle is reduced and local texture matching is triggered; based on the error propagation chain matrix, the multi-vehicle pose data is fused through a weighted averaging algorithm to suppress the error of a single vehicle; the fused pose is fed back to each vehicle node to update the initial state of the inertial navigation.

[0012] Furthermore, the operation plan of the mobile equipment in the port scheduling system is analyzed, and the construction logic of the spatiotemporal reachability prediction model is constructed in combination with the motion state of the corrected vehicle posture. The following steps are used: the operation schedule and path planning data of the mobile equipment in the port scheduling system are extracted, including the movement trajectory of the gantry crane arm and the temporary equipment deployment period; based on the corrected posture data, a vehicle kinematic model is constructed to predict the spatiotemporal area that the vehicle can reach in the future period, and the predicted vehicle trajectory is intersected with the spatiotemporal distribution of the mobile equipment anchor points to generate a spatiotemporal reachability probability distribution map; based on the spatiotemporal density of the intersection area, an anchor point activation priority list is generated, and the anchor points with high probability of being reached are marked.

[0013] Furthermore, the specific process of predicting the spatiotemporal distribution matrix of future reachable anchor points, generating control instructions based on the anchor point activation priority in the matrix, and feeding them back to the dynamic anchor point collaborative construction module is as follows: quantizing the spatiotemporal reachability probability distribution map into a matrix with the matrix dimension of time-space-anchor point identifier, sorting the spatiotemporal nodes in the matrix according to their probability values, generating anchor point activation priority instructions, and giving priority to waking up anchor points with a high probability of being reached; feeding back the priority instructions to the dynamic anchor point collaborative construction module through the port wireless private network, triggering the anchor point wake-up or sleep operation, and monitoring the anchor point activation effect in real time. If the vehicle fails to reach the anchor point as predicted, the priority instructions are dynamically adjusted and the matrix is regenerated.

[0014] A method for positioning a fleet inside and outside a port includes the following steps: S1. Real-time acquisition of dynamic position data of mobile equipment in the port, dynamic calculation of the anchor signal coverage range based on the signal propagation attenuation characteristics in a metal environment, and generation of anchor activation instructions and anchor coverage blind area information in combination with cluster analysis of vehicle historical trajectories; S2. Receive anchor coverage blind area information, perform local grayscale entropy segmentation on ground images captured by on-board cameras, screen high-resolution texture areas, eliminate dynamic interference through cross-frame Bayesian probability association, and generate pose estimation data that is temporally and spatially aligned with the anchor signal; S3. Combine pose estimation data and vehicle inertial navigation data with the communication topology relationship between vehicles to construct an error propagation chain model, and correct the accumulated positioning error through a multi-vehicle confidence weight fusion strategy; S4. Analyze the operation plan of mobile equipment in the port scheduling system, and construct a spatiotemporal accessibility prediction model in combination with the motion state of the corrected vehicle pose, predict the spatiotemporal distribution matrix of future reachable anchor points, and generate control instructions based on the anchor activation priority in the matrix and feed them back to step S1.

[0015] The present invention has the following beneficial effects:

[0016] (1) The internal and external fleet positioning system of the port uses a dynamic anchor point collaborative construction module to dynamically calculate the anchor point signal coverage range based on the signal propagation attenuation characteristics in a metal environment. Combined with the cluster analysis of vehicle historical trajectories, it generates anchor point activation instructions and blind spot information, effectively solving the coverage blind spot problem caused by signal multipath reflection in the dense metal scene of the port, and improving the anchor point resource utilization and positioning continuity. Through the anti-interference environment feature extraction module, local grayscale entropy segmentation and cross-frame Bayesian probability association are performed on the ground image within the coverage blind spot, screening high-resolution texture features and eliminating dynamic interference (such as container projections and water reflections), significantly improving the temporal and spatial alignment accuracy of visual positioning and anchor point signals, and ensuring the robustness of positioning data in complex environments.

[0017] (2) A method for positioning fleets inside and outside the port. This method constructs an error propagation chain model based on the dynamic communication topology between vehicles. It combines a multi-vehicle confidence weight fusion strategy to correct the accumulated positioning error, breaking through the bottleneck of long-term error accumulation in single-vehicle inertial navigation and improving the overall positioning stability of the fleet. The port dispatching system's operation plan is analyzed and a spatiotemporal reachability prediction model is constructed based on the vehicle's motion status. Anchor activation priority instructions are dynamically generated and closed-loop fed back to the anchor coordination construction module to achieve on-demand scheduling and predictive response of anchor resources, significantly reducing the positioning system's reliance on fixed infrastructure while improving the dispatching efficiency and operational safety of the port's temporary fleet.

[0018] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of the port internal and external fleet positioning system of the present invention.

[0020] Figure 2 This is a flow chart of the method for positioning fleets inside and outside a port according to the present invention. DETAILED DESCRIPTION

[0021] The embodiments of the present application use a system and method for positioning fleets inside and outside the port to solve the problems of frequent vehicle positioning blind spots, sensitivity to dynamic interference, uncontrollable error accumulation, and inefficient anchor resource scheduling in complex port environments.

[0022] The overall idea of the solution in the embodiments of this application is as follows:

[0023] Collaborative construction of dynamic anchor points: By acquiring the dynamic location data of mobile equipment in the port in real time, combined with the metal environment signal attenuation correction model and vehicle historical trajectory cluster analysis, the anchor point activation instructions and coverage blind spot information are dynamically generated to solve the problems of high cost of fixed anchor point deployment and random signal coverage blind spots.

[0024] Interference-resistant environmental feature extraction: Within coverage blind spots, high-resolution ground textures are screened based on local grayscale entropy segmentation. Cross-frame Bayesian probabilistic association is used to eliminate instantaneous interference (such as container shadows and reflections from accumulated water). This generates pose data that is temporally and spatially aligned with the anchor point signal, eliminating the reliance of visual positioning on manual labeling.

[0025] Team collaboration error suppression: An error propagation chain model is constructed based on the dynamic communication topology relationship between vehicles. The accumulated inertial navigation error is corrected through a multi-vehicle confidence weight fusion strategy to solve the problem of long-term positioning drift of a single vehicle and improve the overall positioning consistency of the team.

[0026] Anchor point spatiotemporal prediction and closed-loop control: Analyze the port dispatching system's operation plan, build a spatiotemporal reachability prediction model based on vehicle motion status, dynamically generate anchor point activation priority instructions and feed them back to the anchor point collaboration module, enabling on-demand scheduling and predictive response of anchor point resources, reducing hardware dependency costs.

[0027] See also Figure 1 The embodiment of the present invention provides a technical solution: a port internal and external fleet positioning system, including the following modules: a dynamic anchor point collaborative construction module, an anti-interference environment feature extraction module, a fleet collaborative error suppression module, and an anchor point spatiotemporal prediction module; the dynamic anchor point collaborative construction module is used to obtain the dynamic position data of the port mobile equipment in real time, dynamically calculate the anchor point signal coverage range according to the signal propagation attenuation characteristics under the metal environment, and generate the anchor point activation instruction and anchor point coverage blind area information in combination with the vehicle historical trajectory cluster analysis; the anti-interference environment feature extraction module is used to receive the anchor point coverage blind area information, perform local grayscale entropy segmentation on the ground image collected by the vehicle-mounted camera, and screen the high-resolution Texture area, and eliminate dynamic interference through cross-frame Bayesian probability association to generate pose estimation data that is spatiotemporally aligned with the anchor point signal; the fleet collaboration error suppression module is used to receive pose estimation data and vehicle inertial navigation data, build an error propagation chain model based on the communication topology relationship between vehicles, and correct the accumulated positioning error through the multi-vehicle confidence weight fusion strategy; the anchor point spatiotemporal prediction module is used to analyze the operation plan of mobile equipment in the port scheduling system, build a spatiotemporal reachability prediction model based on the motion state of the corrected vehicle pose, predict the spatiotemporal distribution matrix of future reachable anchor points, and generate control instructions based on the anchor point activation priority in the matrix and feed them back to the dynamic anchor point collaboration construction module.

[0028] In this implementation, the dynamic anchor point collaborative construction module is used to dynamically establish an anchor point network based on the real-time position changes of mobile equipment within the port (such as automated guided vehicles (AGVs) and container trucks). Its goal is to improve the spatial distribution rationality and dynamic adaptability of positioning signals, avoiding signal redundancy and blind spots. Dynamic location data refers to the real-time location information uploaded by mobile equipment within the port (which can be obtained through GNSS, inertial navigation, or integrated navigation). Signal propagation attenuation characteristics: In metal-dense environments, electromagnetic wave propagation experiences significant attenuation and multipath effects. This method takes this characteristic into account to dynamically estimate the effective signal coverage range of anchor points. Anchor point signal coverage refers to the area within which a single anchor point device can stably provide positioning services under current environmental conditions. Historical vehicle trajectory clustering analysis: By clustering historical positioning trajectories, common paths and high-frequency operation areas are extracted, providing a statistical basis for anchor point deployment and activation decisions. Anchor point activation instructions and blind spot information: The former indicates which anchor points should be activated to provide positioning services in the current or future time periods; the latter marks areas where the current anchor point signal is unreachable. Interference-Resistant Environmental Feature Extraction Module: This module uses the vehicle's visual perception system to assist in positioning within anchor signal blind spots, ensuring stable vehicle pose estimation even when anchors are unavailable. Local Grayscale Entropy Segmentation: Based on the local entropy of the grayscale distribution within an image, it extracts texture regions with rich information and high discrimination, helping to improve the stability and uniqueness of image features. Grayscale Entropy: This reflects the degree of disorder (information entropy) of the grayscale distribution within an image and is used to determine region complexity. Highly Discriminative Texture Regions: Image regions with clear boundaries, distinct structures, and low repetitiveness are suitable for feature matching and pose estimation. Cross-Frame Bayesian Probabilistic Association: This module uses Bayesian inference to perform probabilistic matching of texture features across time-series images to eliminate feature points affected by interference factors such as occlusion and dynamic objects. Bayesian Probabilistic Association: This module updates the matching confidence level based on a prior distribution and observed data, making it a highly robust dynamic feature matching method. Pose Estimation: This module represents the vehicle's position and pose (including position coordinates and heading angle) at a specific moment in time and is used for navigation and path tracking. Fleet collaborative error suppression module: This module is aimed at multi-vehicle collaborative positioning applications. By introducing communication topology modeling and confidence fusion strategies, it enables the sharing and complementarity of positioning data between vehicles, and suppresses the accumulation and diffusion of single-vehicle positioning errors. Inertial navigation data: Short-term position change information calculated based on inertial components such as accelerometers and gyroscopes, but there are long-term drift errors. Communication topology relationship: Describes the information transmission structure between vehicles in the fleet, such as star, ring or mesh structure, which determines the path along which error information can be propagated. Error propagation chain model: Models how error information propagates in a fleet, and is used to estimate the potential error transmission path and superposition effect of each vehicle.Multi-vehicle confidence weight fusion strategy: Dynamically assigns confidence weights based on the stability of each vehicle's positioning data, sensor quality, and communication stability, performing distributed filtering fusion to correct for inter-vehicle differential errors. Confidence weight: A numerical indicator used to reflect the credibility of the data source. Anchor point spatiotemporal prediction module: This module combines the work plan in the port scheduling system to predict the spatiotemporal distribution of future reachable anchor points, thereby optimizing the anchor point activation strategy in advance, enabling proactive deployment of positioning services and dynamic scheduling of system resources. Work plan parsing: This module reads data from the scheduling system regarding future equipment work tasks, such as path planning and loading and unloading arrangements. Motion state modeling: This module constructs a future motion trajectory model based on parameters such as the vehicle's current position, velocity, and acceleration. Spatiotemporal reachability prediction model: This module integrates the work plan and motion model to predict the areas that a vehicle may pass through within a certain time period and the spatial range that can be covered by the anchor points. Anchor point activation priority: Based on the prediction model, each candidate anchor point is ranked by its importance to future positioning tasks, which is used to indicate the anchor point scheduling priority. Control instruction feedback: The priority decision results are fed back to the dynamic anchor point module to form a task-driven closed-loop anchor point optimization system.

[0029] Specifically, the dynamic position data of the mobile equipment in the port is acquired in real time, and the specific process of dynamically calculating the anchor point signal coverage range according to the signal propagation attenuation characteristics in the metal environment is as follows: the dynamic position data of the mobile equipment in the port is analyzed in real time to obtain the real-time coordinates and movement direction of the gantry crane boom and temporary traffic equipment; according to the distribution density of metal obstacles in the container stacking area, the attenuation coefficient of the signal propagation model is adjusted in segments to generate a non-uniform attenuation gradient; based on the corrected attenuation gradient, the effective coverage boundary of the signal is calculated anchor by anchor point, and a vector map described by polygon vertex coordinates is generated; the real-time position of the vehicle is geometrically matched with the vector map, and the areas with signal strength lower than the positioning threshold are marked as coverage blind spots, and a list of blind spot vertex coordinates is output.

[0030] In this implementation plan, the steps for real-time analysis of the dynamic location data of mobile equipment in the port are as follows: The dynamic location information of mobile equipment in the port is obtained in real time through the vehicle-mounted positioning terminal (GNSS / IMU) and the positioning gateway deployed in the port dispatching system. The analyzed data contains the following information: vehicle number and type (AGV, tractor, empty container truck, etc.); current coordinate point; movement direction (unit vector form); and heading angle ); the location and movement trajectory of other dynamic facilities (such as gantry crane arms, guide vehicles, and temporary obstacles). Analyze the distribution density of metal obstacles and adjust the signal propagation model. Steps: Use the port 3D modeling system or LiDAR point cloud reconstruction model to obtain the spatial distribution characteristics of metal objects such as container stacking areas and crane structures. Based on the density of metal objects and the obstruction structure, correct the attenuation of the signal propagation path. Non-uniform attenuation modeling: Based on the density distribution of metal obstacles in the path, establish a set of segmented corrected signal attenuation functions. Assume: : The propagation power of the j-th anchor point signal in the i-th path; : the distance of the path segment; : The environmental attenuation factor related to the metal shielding density of this section. Then the signal power of this section can be expressed as: ;in: : the initial transmit power of the jth anchor point; : Adjust according to the on-site obstruction situation, such as setting it to 3.0-4.5 in a high-density container area and 2.0-2.5 in an open area. The overall signal attenuation result from the anchor point j to any coordinate point M can be obtained by summing multiple paths. Calculate the anchor point signal coverage boundary step description: Based on the port's two-dimensional geographic coordinate grid, start from each anchor point to scan in all directions. According to the above non-uniform model, determine the signal strength on the ray path to drop to the threshold. The farthest point is used as the boundary point in that direction. Polygon boundary construction method: connect the boundary points in each direction to form an irregular polygon boundary representing the valid area of the anchor point signal. The polygon boundary is represented in the form of a vertex coordinate list: ;in, For the The signal coverage area of the anchor point, is the number of boundary vertices. Steps for building a vector map and matching vehicle positions: Based on the anchor polygon boundary, build a real-time vector map of the port and match the current position of each vehicle. Project it onto the map. Use the point-in-polygon algorithm (such as the ray method, the winding number method, etc.) to determine whether the vehicle is within the effective coverage area of a certain anchor point: If , then the vehicle is subject to Anchor signal coverage; if the vehicle is not in any The combined signal power within or at that point , then the point is judged as an "anchor blind area". Mark the blind area and output the blind area polygon coordinates. Step description: All areas not covered by any anchor point valid signal are spatially clustered to extract continuous blind area boundaries. Use DBSCAN density clustering, for example , output its vertex coordinate list: ;in, Indicates the These coordinates will serve as the input basis for subsequent modules (such as the anti-interference image perception module) to guide them to perform visual-assisted positioning in the blind spot area.

[0031] Specifically, the specific process of generating anchor point activation instructions and anchor point coverage blind spot information in combination with vehicle historical trajectory clustering analysis is as follows: collect vehicle historical trajectory data, and identify high-frequency traffic areas, including container transfer areas and gate channels, through density clustering algorithm; generate anchor point activation priority instructions based on clustering results, and give priority to activating mobile device anchor points in high-frequency areas; match the coverage blind spot coordinates with the vehicle's real-time position. When the vehicle enters the blind spot boundary, activate the adjacent mobile device anchor points according to the blind spot vertex coordinates, and extend their working time. Update the anchor point activation timing table according to the vehicle's real-time position to adjust the signal coverage range.

[0032] In this implementation plan, the steps for collecting and preprocessing vehicle historical trajectory data are as follows: Retrieve multiple days of vehicle trajectory logs from the port dispatching system. The trajectory data format is as follows: ;in: :No. Historical trajectory data of the vehicle; :No. The location at a point in time; : corresponding timestamp; : The total number of track points recorded by the vehicle. After all vehicle track sets are time-aligned and spatially normalized, they are fed into the clustering model. 2. Identifying High-Frequency Traffic Areas Based on Density Clustering Algorithm Step Description: Use a density-based spatial clustering algorithm (such as DBSCAN) to process the track point set to identify high-frequency vehicle activity areas. Suppose the track set is: ; The output after clustering is: ;in: :No. A high-density traffic cluster; : The number of trajectory points contained in the cluster; the cluster will correspond to a specific port area scenario, such as the main channel of the yard, the gate entrance and exit. Generate anchor activation priority instruction step description: Quantitatively analyze the clusters according to vehicle traffic frequency, time period distribution, and traffic density, and calculate the priority score λ_q for each cluster. The scoring model is: ;in: : Frequency of vehicle traffic within the cluster; : average vehicle density per unit time; : Time coverage of the intra-cluster traffic period; : Weight coefficient (set according to the port area operation weight). Sort from high to low to generate an anchor priority activation list , guiding the dynamic anchor point collaborative construction module to prioritize the mobile anchor points corresponding to the high-traffic area. Dynamically activate adjacent anchor points based on blind spot information. Step description: Based on the blind spot polygon vertex list identified in the previous steps: ; The vehicle's current position The Euclidean distance is determined with the blind spot boundary, and the nearest point of the blind spot boundary is , then: ;when (Set distance threshold), if the vehicle is about to enter the blind spot, then: send an early activation command to the mobile device anchor point near the blind spot boundary; according to the vehicle's speed , estimated residence time in the blind spot , set the anchor delay closing time to the current time plus Dynamically update the anchor point activation schedule and coverage step description: Establish the anchor point control schedule , represents the activation status of each anchor point at time t: ;in: : the sth anchor point number; The estimated closing time is dynamically updated based on the vehicle's current location, predicted path, and blind spot dwell time. The activation status is determined by whether the vehicle is within the blind spot boundary and clustered in a high-priority zone. The anchor point coverage priority and each anchor point's operating hours are continuously updated based on the vehicle's trajectory, ensuring that blind spots are not left uncovered for extended periods and preventing energy waste caused by redundant anchor point activation.

[0033] Specifically, the ground image captured by the vehicle-mounted camera is subjected to local grayscale entropy segmentation, and the specific process of screening high-resolution texture areas is as follows: within the coverage blind area, the ground image captured by the vehicle-mounted camera is divided into multiple square local blocks, the local grayscale entropy value of each block is calculated, and the blocks with grayscale entropy values higher than the environmental background are screened as candidate texture areas. The entropy value changes of the same blocks in adjacent frame images are compared, and abnormal high-entropy blocks caused by instantaneous reflection or projection are eliminated; morphological closing operations are performed on the candidate texture areas, the broken areas are filled, and a continuous positioning reference mask is generated.

[0034] In this embodiment, the steps of image area division and local grayscale entropy calculation are as follows: the grayscale image captured by the vehicle camera is Divide into multiple non-overlapping local square blocks: ;in: : The ground grayscale image of the current frame; :No. Rank Columns of local image blocks; : The grid dimension of the divided blocks is determined by the image size and block granularity. Local grayscale entropy calculation: For each block Calculate its grayscale entropy : ;in: : Grayscale level (usually 256); : Gray value is Pixels in the block Normalized frequency in ; : To prevent the logarithmic term from appearing Small constants (such as ). Candidate texture block screening and high entropy anomaly removal steps: 1. Background entropy threshold setting: Calculate the entropy mean of all blocks in the entire image and standard deviation , set the background entropy threshold: ;in: : Adjustment coefficient, control the screening intensity (usually 1~2); the area exceeding this threshold is considered as a candidate block with texture differentiation. 2. Inter-frame abnormal high entropy elimination: For two consecutive frames of images and , compare the entropy difference of blocks at the same position: ;like , then the change is judged to be a non-structural disturbance (such as reflection, projection), and the block is eliminated. : Set a threshold to reflect the maximum entropy change of stable texture (experienced value such as 0.5~1.0); after elimination, retain stable and repeatable texture blocks. 3. Morphological closing operation repair and mask generation steps: Binary mask map of candidate texture blocks Perform morphological closing operations (dilation followed by erosion) to eliminate small holes and connect broken areas: ; Where: Dilate, Erode: represent image dilation and erosion operations respectively; : Structural elements (such as Square kernel); the result of the operation is a closed and continuous positioning reference mask image. Output continuous texture positioning mask final result: Output This mask is used as the texture region suitable for positioning matching in the frame image. This mask is then spatially aligned with the anchor signal coverage information and fed into the subsequent cross-frame Bayesian pose estimation module. This ensures that the vehicle can still rely on stable ground texture for auxiliary positioning in blind spots, improving positioning robustness.

[0035] Specifically, dynamic interference is eliminated through cross-frame Bayesian probability association to generate pose estimation data that is temporally and spatially aligned with the anchor signal. The specific process is as follows: stable texture features in consecutive multi-frame images are temporally and spatially calibrated, the motion trajectory of feature points is recorded, a cross-frame feature association probability matrix is constructed, and the association probability between the current frame feature and the historical frame is calculated based on the Bayesian inference model. If the probability is lower than the dynamic interference threshold, the feature is eliminated; the filtered features are aligned with the anchor signal timestamp, the vehicle pose is solved through the perspective transformation matrix, and the multi-frame pose estimation results are fused to generate smoothed vehicle trajectory data.

[0036] In this implementation, the steps of temporal and spatial calibration of stable texture features and feature trajectory construction are as follows: A set of key feature points of the stable texture area is extracted from the mask image that has completed grayscale entropy segmentation and morphological processing: ;in: : Image frame time index; :No. The image coordinates of feature points; :time The texture feature point set extracted at each moment. Use the optical flow method or feature matching algorithm to track the same points in multiple consecutive frames and construct the motion trajectory sequence of each feature point: Cross-frame feature association probability modeling and Bayesian inference steps: For each pair of feature point trajectories, based on factors such as their image spatial position similarity, motion direction consistency, and time interval attenuation factor, construct the cross-frame feature association probability: ;in: :No. Features in Moment and Features in Bayesian association probability at time; : Likelihood function, measuring the difference between position and motion; : Prior probability, set according to the stability of regional texture; the denominator is a normalization factor to ensure that the sum of the association probabilities of all candidate feature points is 1. Note: The likelihood function can be defined as a two-dimensional Gaussian distribution. The specific formula is omitted and can be found in the literature on image matching. If the maximum association probability of a feature point is lower than the dynamic interference threshold , then the point is determined to be a dynamic interference source (such as a pedestrian or a robotic arm) and is removed: . Description of the steps for time-space alignment with anchor point signals and vehicle posture solution: Time synchronization: For the retained stable feature points, Time cut with anchor signal Alignment, using linear interpolation or nearest neighbor matching for time fusion. Single frame perspective transformation matrix solution: using the selected plane feature points and their corresponding physical coordinates on the real port ground, through the homography matrix Establish a mapping relationship between the image and the world coordinates: ;in: : Homogeneous coordinates of feature points in the image coordinate system; : corresponds to the position in world coordinates; : The perspective transformation matrix estimated based on the matching point pairs; Reverse the two-dimensional pose of the vehicle body . Multi-frame pose fusion to generate trajectory steps: Estimated vehicle pose for each frame ,The weighted filtering strategy in the sliding window is used to smooth the trajectory: ;in: : Smoothing weights, which give higher weights to the center frame (such as Gaussian weights or triangular kernels); : Sliding window radius; : Smoothed pose after filtering; continuous The composed trajectory is used for subsequent error suppression module processing.

[0037] Specifically, the model receives pose estimation data and vehicle inertial navigation data and constructs an error propagation chain model based on the inter-vehicle communication topology. The model dynamically establishes a mesh communication topology based on inter-vehicle wireless communication signal strength and distance thresholds. A table of vehicle node connections is maintained in real time. The accumulated inertial navigation errors of individual vehicles are mapped to directed graph nodes, and error propagation weights are constructed based on the pose differences between adjacent vehicles. Based on the consistency of vehicle motion directions, the error transmission attenuation coefficient along the communication link is calculated to generate an error propagation chain matrix. The error propagation weights and attenuation coefficients are recalculated when the vehicle position or communication topology changes.

[0038] In this implementation plan, the communication topology is constructed and the connection status is maintained. The dynamic communication topology is constructed based on the wireless communication signal strength (RSSI) and the maximum communication distance threshold between the vehicles in the fleet. , dynamically build a mesh communication topology ,in: : represents all vehicle nodes in the fleet; : represents the set of vehicle pairs that meet the communication conditions; communication edge If and only if: ;in: :vehicle and The Euclidean distance of :The lower threshold of communication signal strength. Node connection status table updates. Each vehicle node broadcasts status heartbeat packets regularly; the central controller or edge node summarizes the communication status and generates a topology connection status table. ,in: The single vehicle error node represents the calculation of adjacent error weights. The single vehicle error node defines the cumulative inertial navigation error of each vehicle. Expressed as the node error value in the graph, where: , represents the position residual between the inertial solution and the visual pose; :vehicle The current pose calculated by inertial navigation; :vehicle The pose calculated by vision / anchor alignment. The neighboring error propagation weights are for any pair of neighboring vehicles ( ), define the error propagation weight Indicates the error from the vehicle Transfer to vehicle Credibility: ;in: : Error amplification factor (set according to system experience); : represents the Euclidean distance between the visually estimated poses; the larger the weight value, the closer the poses of the two vehicles are, and the error can be effectively propagated and compensated. The motion direction consistency and error propagation attenuation modeling, the direction consistency factor is set as vehicle 、 The motion direction vectors are: ; Then the directional consistency factor of the two is defined as: ; If the two vehicles are heading in the same direction (collinear and in the same direction), If the direction is opposite, The error attenuation coefficient is calculated by comprehensively considering the direction consistency and communication strength, and the error attenuation coefficient is defined as: ; If the vehicle directions are inconsistent ( ), then the attenuation coefficient is 0; the attenuation coefficient of the effective link determines the efficiency of error transmission within the fleet. Error propagation chain matrix construction and dynamic update mechanism, error propagation chain matrix definition Error propagation chain matrix: ; Its elements are: ; Each row represents a vehicle The attenuation intensity of the error that may be propagated to other vehicles; error tracing, weighted feedback and fusion can be achieved on the matrix diagram. Dynamic recalculation mechanism When the position change of any vehicle exceeds the threshold , or the communication topology changes (addition / disconnection of edges), re-execute the following two steps: Rebuild the communication graph Update the connection status table , , ,Reconstruction .

[0039] Specifically, the specific process of correcting the cumulative positioning error through the multi-vehicle confidence weight fusion strategy is as follows: according to the alignment between the vehicle pose estimation data and the anchor point signal and the historical positioning stability, the confidence weights are dynamically allocated; if the pose difference between adjacent vehicles exceeds the set threshold, the weight of the abnormal vehicle is reduced and local texture matching is triggered. Based on the error propagation chain matrix, the multi-vehicle pose data is fused through a weighted averaging algorithm to suppress the error of a single vehicle; the fused pose is fed back to each vehicle node to update the initial state of the inertial navigation.

[0040] In this implementation, the confidence weight distribution logic and dynamic confidence index calculation are used for any vehicle node in the fleet. , let its current visual pose estimate be , the inertial navigation pose is , the anchor signal matching degree is , the historical stability score is , then the vehicle confidence weight Defined as follows: ;in: :Indicates vehicle The degree of alignment between the current pose and the anchor signal; : Indicates the positioning stability of the vehicle in the recent period (such as the inverse of the position fluctuation variance); :Indicates vehicle The set of neighbor vehicles under the communication topology; the denominator is normalized to ensure Note: This formula reflects the weight evaluation mechanism based on the anchor point matching quality and historical performance, which avoids the pollution of the overall positioning by a single abnormal node. Anomaly detection and local texture compensation trigger, posture difference detection for any adjacent vehicle 、 , calculate the current visual pose difference as: ; if present: ; then it is determined or There is an anomaly in the process, triggering the following actions: temporarily reduce the confidence of the abnormal vehicle node (for example, multiply by an adjustment factor ); initiate the local texture relocation process of the abnormal node and perform texture matching verification on the latest frames of images; if the relocation is valid, restore the confidence level, otherwise continue to reduce the influence weight of the node in the fusion model. Based on the weighted fusion correction of the propagation chain matrix, the multi-vehicle posture fusion model is more accurate for the target vehicle. , and its neighborhood vehicle set is , the error propagation chain matrix is , the pose estimation after fusion The calculation is as follows: ;in: : attenuation coefficient in the error propagation chain (see previous section for details); : Confidence based on anchor point matching and historical stability assessment; : Neighboring car The current visual pose of multiple vehicles can significantly suppress the drift error or local failure of a vehicle after weighted fusion. The balanced normalization strategy is to avoid the imbalance of the sum of weights and perform normalization before fusion: The pose feedback and inertial navigation state reinitialization will be the fused smooth pose As a high-confidence reference result; feedback to the vehicle The local navigation system is used to update the initial state of inertial navigation. The specific update formula is as follows: ;in: : Initialize the updated inertial navigation position; : Inertial navigation velocity vector estimation; : The time interval between two navigation status updates.

[0041] Specifically, the operation plan of mobile equipment in the port scheduling system is analyzed, and the construction logic of the spatiotemporal reachability prediction model is constructed in combination with the motion state of the corrected vehicle posture. The following steps are used: the operation schedule and path planning data of the mobile equipment in the port scheduling system are extracted, including the movement trajectory of the gantry crane arm and the temporary equipment deployment period; based on the corrected posture data, a vehicle kinematic model is constructed to predict the spatiotemporal area that the vehicle can reach in the future period, and the predicted vehicle trajectory is intersected with the spatiotemporal distribution of the mobile device anchor points to generate a spatiotemporal reachability probability distribution map; based on the spatiotemporal density of the intersection area, an anchor point activation priority list is generated, and the anchor points with high probability of being reached are marked.

[0042] In this implementation, the operation plan and path planning data are parsed, and the operation schedule is extracted to obtain the operation schedule of mobile equipment (such as gantry crane arms and temporary transportation equipment) from the port scheduling system. The time interval set is defined as follows: ;in Indicates the start and end time nodes of the operation. Path planning data acquisition includes a set of moving trajectory points of the equipment: ;in Represents the position coordinates in three-dimensional space. Vehicle kinematic model construction 1. Input the corrected posture data to the vehicle , with the corrected pose data As the starting point, establish the state vector: ;in: : Vehicle two-dimensional position coordinates; : vehicle heading angle; : Vehicle speed. The kinematic prediction formula uses the vehicle kinematic model to predict the future time Location: ;in: : Vehicle angular velocity (estimated from historical data); : Prediction time step. Generate a spatiotemporal reachability probability distribution map and construct a predicted trajectory using a kinematic model to generate a set of predicted trajectories for the vehicle in the future period [t, t+T]: ; Distribution of spatiotemporal anchor points Let the spatiotemporal position set of mobile device anchor points be: ;in is the anchor point space coordinate, The time interval for activating the anchor point is calculated by the time-space intersection operation. The predicted trajectory of the vehicle is calculated with the time-space distribution of the anchor point. Reach anchor point Probability , combining the spatial distance and time overlap relationship: ;in: represents the Euclidean distance; : Spatial position tolerance radius, reflecting the allowable error range; is the characteristic function, when The spatiotemporal accessibility probability distribution graph integrates all anchor points and prediction periods to generate a probability matrix: The anchor activation priority list is generated, and the spatiotemporal density is calculated based on the spatiotemporal probability matrix , calculate the total probability of reaching the anchor point in the prediction period: ; Priority sorting by Sort from largest to smallest to generate an anchor activation priority list: The high probability anchor mark is for the The anchors of , marked as “high priority activation” anchors, serve as the control input of the dynamic anchor collaborative building module.

[0043] Specifically, the specific process of predicting the spatiotemporal distribution matrix of future reachable anchor points, generating control instructions based on the anchor point activation priority in the matrix, and feeding back the control instructions to the dynamic anchor point collaborative construction module is as follows: quantizing the spatiotemporal accessibility probability distribution map into a matrix with the matrix dimension of time-space-anchor point identifier, sorting the spatiotemporal nodes in the matrix according to their probability values, generating anchor point activation priority instructions, and giving priority to waking up anchor points with a high probability of being reached; feeding back the priority instructions to the dynamic anchor point collaborative construction module through the port wireless private network, triggering the anchor point wake-up or sleep operation, and monitoring the anchor point activation effect in real time. If the vehicle fails to reach the anchor point as predicted, the priority instructions are dynamically adjusted and the matrix is regenerated.

[0044] In this implementation, the spatiotemporal accessibility probability distribution map is quantified into a matrix, and the probability distribution of port vehicles reaching each anchor point in the future within the spatiotemporal range is converted into a three-dimensional matrix representation. The matrix dimensions include time series, spatial region, and anchor point identifier. Let this matrix be ,in: Represents a discrete set of time nodes, the number of which is ; Represents a set of spatial segmentation regions, the number of which is ; Represents the anchor point set, the number is ; Matrix elements Indicates at a point in time , spatial area The vehicle inside reaches the anchor point probability. , calculate the anchor activation priority, based on the matrix The comprehensive reach probability of each anchor point in the calculation of the anchor point Activation priority indicator : ;in: Representing spatiotemporal nodes The weight coefficient reflects the priority or importance of the spatiotemporal node (for example, the weight of the node close to the current time is larger, and the weight of the key operation area is larger); It is an anchor point The anchor activation priority list is based on Sort in descending order, activate first Generate and issue anchor control instructions based on the activation priority threshold , for each anchor point Generate activation control instructions: ;in: Indicates the activation status of the anchor point, 1 is active and 0 is dormant; threshold It can be dynamically adjusted according to system load and resource conditions. It sends control instructions to the port wireless private network and feeds back to the dynamic anchor point collaborative construction module to execute the corresponding anchor point wake-up or sleep operation. The real-time monitoring and dynamic adjustment system continuously monitors the anchor point activation effect and records the actual situation of the vehicle reaching the anchor point. If the vehicle does not reach the activation anchor point as predicted, the activation priority weight or threshold is adjusted according to the actual data, and the spatiotemporal distribution matrix is regenerated. And update the activation instructions. This process can be represented by the following correction mechanism: ;in: Predict the probability for the previous period; It is a probability estimate updated based on actual monitoring data; is the smoothing coefficient, which weighs historical predictions against real-time adjustments.

[0045] See also Figure 2, a method for positioning a fleet inside and outside the port, including the following steps: S1. Real-time acquisition of dynamic position data of mobile equipment in the port, dynamic calculation of the anchor signal coverage range according to the signal propagation attenuation characteristics in the metal environment, and generation of anchor activation instructions and anchor coverage blind area information in combination with vehicle historical trajectory clustering analysis; S2. Receive anchor coverage blind area information, perform local grayscale entropy segmentation on the ground image captured by the on-board camera, screen high-resolution texture areas, and eliminate dynamic interference through cross-frame Bayesian probability association to generate pose estimation data that is time-space aligned with the anchor signal; S3. Combine the pose estimation data and vehicle inertial navigation data with the communication topology relationship between vehicles to construct an error propagation chain model, and correct the accumulated positioning error through the multi-vehicle confidence weight fusion strategy; S4. Analyze the operation plan of the mobile equipment in the port scheduling system, and construct a spatiotemporal reachability prediction model in combination with the motion state of the corrected vehicle pose, predict the spatiotemporal distribution matrix of the future reachable anchor points, and generate control instructions based on the anchor activation priority in the matrix and feed it back to step S1.

[0046] In this implementation, step S1: Real-time adjustment of the dynamic anchor point signal coverage breaks through the traditional assumption of fixed anchor point coverage. It dynamically models signal attenuation characteristics in metal-dense environments, accurately characterizes non-uniform signal propagation, and enables real-time dynamic calculation of coverage, significantly improving the spatial effectiveness of positioning signals. Cluster analysis, integrated with vehicle historical trajectories, innovatively utilizes density clustering to identify high-frequency traffic areas. This allows for spatiotemporal priority scheduling of anchor point activations, effectively optimizing resource allocation, avoiding invalid activations, and improving system energy efficiency and response speed. A blind spot information feedback mechanism enables precise positioning of coverage blind spots. It dynamically activates neighboring anchor points based on the coordinates of blind spot vertices, compensating for signal blind spots and enhancing the continuity and stability of the positioning system. Step S2: Local grayscale entropy segmentation of the positioning reference area uses a local grayscale entropy method to select high-resolution textures, improving the stability and recognition rate of visual features and effectively addressing lighting and texture variations in the complex port environment. A cross-frame Bayesian probabilistic association model incorporates a Bayesian inference mechanism to verify the spatiotemporal consistency of texture features across consecutive frames. This innovatively eliminates dynamic interference factors such as reflections and shadows, ensuring the accuracy and robustness of pose estimation data. A spatiotemporal alignment fusion mechanism synchronizes visual features with anchor signal timestamps, integrating multimodal information and improving overall positioning accuracy. Step S3: A dynamic communication topology based on wireless communication signal strength reflects the vehicle network connection status in real time, dynamically establishing a multi-vehicle error propagation chain, breaking through the limitations of single-vehicle positioning and enabling cross-vehicle error propagation and control. The innovative error propagation chain model maps the accumulated inertial navigation errors into directed graph nodes. The error propagation weights and attenuation coefficients are calculated using communication topology and motion consistency, scientifically revealing the error propagation paths and impact. A multi-vehicle confidence weighted fusion strategy dynamically allocates vehicle positioning confidence. Through weighted fusion, it suppresses anomalous vehicle errors, improves the stability and reliability of overall fleet positioning, and achieves collaborative error correction. Step S4: The deep integration of scheduling operation plans and motion status integrates the port scheduling system operation plan into the positioning prediction process for the first time. Combined with a vehicle kinematic model with corrected posture, it accurately predicts the spatiotemporal area of future anchor point contact, enabling forward-looking positioning resource scheduling. The spatiotemporal reachability prediction model, based on the intersection of trajectory predictions and the spatiotemporal distribution of anchor points, quantifies the probability of reaching anchor points and systematically constructs a spatiotemporal distribution matrix to scientifically guide anchor activation priority. A closed-loop control feedback mechanism generates control commands to adjust the dynamic anchor collaborative construction module in real time, enabling adaptive adjustment of the positioning system and improving its real-time responsiveness and accuracy.

[0047] In summary, this application has at least the following effects:

[0048] A system and method for positioning fleets inside and outside ports combines dynamic anchor point collaborative construction with anti-interference environmental feature extraction to effectively suppress signal propagation attenuation and the effects of dynamic interference in metallic environments, achieving more accurate vehicle pose estimation. Dynamic interference features are eliminated using cross-frame Bayesian probabilistic associations. An error propagation chain model is constructed in conjunction with the inter-vehicle communication topology. A multi-vehicle confidence weight fusion strategy effectively suppresses the cumulative error of single-vehicle inertial navigation, improving the stability and robustness of the overall positioning system. By clustering vehicle historical trajectories and predicting spatiotemporal reachability, dynamic adjustment of anchor point activation is achieved, avoiding inefficient anchor point resource dissipation, optimizing signal coverage, and reducing system energy consumption and maintenance costs. A closed-loop feedback mechanism based on a spatiotemporal distribution matrix and dynamic priority instructions enables real-time adjustment and optimization of the anchor point activation strategy, ensuring the system's rapid response to changes in the port's operating environment. Employing a mesh communication topology and an error propagation chain model, the system integrates multi-vehicle pose data to meet the multi-node collaborative positioning requirements of large-scale port fleets, improving overall fleet efficiency and safety management.

[0049] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, 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 implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0050] The present invention is described with reference to flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0051] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0052] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0053] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0054] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. The port internal and external fleet positioning system is characterized by: It includes the following modules: dynamic anchor point collaborative construction module, anti-interference environment feature extraction module, fleet collaborative error suppression module, and anchor point spatiotemporal prediction module; The dynamic anchor point collaborative construction module is used to obtain the dynamic position data of the port mobile equipment in real time, dynamically calculate the anchor point signal coverage range based on the signal propagation attenuation characteristics in the metal environment, and generate anchor point activation instructions and anchor point coverage blind area information based on the cluster analysis of the vehicle historical trajectory; The anti-interference environment feature extraction module is used to receive anchor point coverage blind spot information, perform local grayscale entropy segmentation on the ground image captured by the vehicle-mounted camera, screen high-resolution texture areas, and eliminate dynamic interference through cross-frame Bayesian probability association to generate pose estimation data that is temporally and spatially aligned with the anchor point signal; The platoon collaborative error suppression module is used to receive pose estimation data and vehicle inertial navigation data, build an error propagation chain model based on the inter-vehicle communication topology, and correct the accumulated positioning error through a multi-vehicle confidence weight fusion strategy; The anchor point spatiotemporal prediction module is used to analyze the operation plan of mobile equipment in the port scheduling system, build a spatiotemporal reachability prediction model based on the motion state of the corrected vehicle posture, predict the spatiotemporal distribution matrix of future reachable anchor points, and generate control instructions based on the anchor point activation priority in the matrix and feed them back to the dynamic anchor point collaborative construction module; The logic for receiving pose estimation data and vehicle inertial navigation data and building an error propagation chain model based on the inter-vehicle communication topology is as follows: Based on the inter-vehicle wireless communication signal strength and distance threshold, a mesh communication topology is dynamically established, and a vehicle node connection status table is maintained in real time. The accumulated error of a single vehicle's inertial navigation is mapped to a directed graph node, and the error propagation weight is constructed based on the posture differences of adjacent vehicles. According to the consistency of the vehicle's motion direction, the error transmission attenuation coefficient along the communication link is calculated to generate the error propagation chain matrix. When the vehicle position or communication topology changes, the error propagation weight and attenuation coefficient are recalculated; The specific process of correcting the cumulative positioning error through the multi-vehicle confidence weight fusion strategy is as follows: Dynamically assign confidence weights based on the alignment between the vehicle pose estimation data and the anchor point signal and the historical positioning stability; If the difference in the posture of adjacent vehicles exceeds the set threshold, the weight of the abnormal vehicle is reduced and local texture matching is triggered. Based on the error propagation chain matrix, the multi-vehicle posture data is fused through a weighted average algorithm to suppress the error of a single vehicle; The fused pose is fed back to each vehicle node to update the initial state of inertial navigation.

2. The port internal and external fleet positioning system according to claim 1, characterized in that: The specific process of obtaining the dynamic position data of the port mobile equipment in real time and dynamically calculating the anchor point signal coverage range based on the signal propagation attenuation characteristics in the metal environment is as follows: Real-time analysis of the dynamic location data of mobile equipment in the port, obtaining the real-time coordinates and movement direction of gantry crane arms and temporary transportation equipment; Based on the distribution density of metal obstacles in the container stacking area, the attenuation coefficient of the signal propagation model is adjusted in sections to generate a non-uniform attenuation gradient. Based on the corrected attenuation gradient, the effective signal coverage boundary is calculated anchor by anchor point, and a vector map described by polygon vertex coordinates is generated. The vehicle's real-time position is geometrically matched with the vector map, and areas where signal strength is below the positioning threshold are marked as coverage blind spots. A list of blind spot vertex coordinates is then output.

3. The port internal and external fleet positioning system according to claim 2, characterized in that: The specific process of generating anchor point activation instructions and anchor point coverage blind spot information based on cluster analysis of vehicle historical trajectories is as follows: Collect historical vehicle trajectory data and use density clustering algorithms to identify high-frequency traffic areas, including container transfer areas and gate channels; Generate anchor point activation priority instructions based on clustering results, giving priority to activating mobile device anchor points in high-frequency areas; Match the coverage blind spot coordinates with the vehicle's real-time position. When the vehicle enters the blind spot boundary, activate the adjacent mobile device anchor point according to the blind spot vertex coordinates and extend its working time. Update the anchor point activation timing table according to the vehicle's real-time position and adjust the signal coverage range.

4. The port internal and external fleet positioning system according to claim 3, characterized in that: The specific process of performing local grayscale entropy segmentation on the ground image captured by the vehicle-mounted camera and screening high-resolution texture areas is as follows: Within the coverage blind spot, the ground image captured by the vehicle-mounted camera is divided into multiple square local blocks. The local grayscale entropy value of each block is calculated. The blocks with grayscale entropy values higher than the environmental background are selected as candidate texture areas. The entropy value changes of the same blocks in adjacent frame images are compared to eliminate abnormally high entropy blocks caused by instantaneous reflection or projection. Perform morphological closing operation on the candidate texture area, fill the broken area, and generate a continuous positioning reference mask.

5. The port internal and external fleet positioning system according to claim 4, characterized in that: The specific process of eliminating dynamic interference by cross-frame Bayesian probability association and generating pose estimation data that is spatially and temporally aligned with the anchor signal is as follows: Perform spatiotemporal calibration on stable texture features in multiple consecutive frames of images, record the motion trajectory of feature points, construct a cross-frame feature association probability matrix, and calculate the association probability between the current frame feature and the historical frame based on the Bayesian inference model. If the probability is lower than the dynamic interference threshold, the feature is eliminated. The filtered features are aligned with the anchor signal timestamps, the vehicle pose is solved using the perspective transformation matrix, and the multi-frame pose estimation results are fused to generate smoothed vehicle trajectory data.

6. The port internal and external fleet positioning system according to claim 5, characterized in that: The logic for analyzing the operation plan of mobile equipment in the port dispatching system and building a spatiotemporal reachability prediction model based on the motion state of the corrected vehicle posture is as follows: Extracting the operation schedule and path planning data of mobile equipment from the port dispatching system, including the movement trajectory of gantry crane arms and the deployment period of temporary equipment; Based on the corrected posture data, a vehicle kinematic model is constructed to predict the spatiotemporal area that the vehicle can reach in the future. The predicted vehicle trajectory is intersected with the spatiotemporal distribution of the mobile device anchor points to generate a spatiotemporal reachability probability distribution map. Based on the spatiotemporal density of the intersection area, a priority list of anchor point activation is generated, marking the anchor points with high probability of being reached.

7. The port internal and external fleet positioning system according to claim 6, characterized in that: The specific process of predicting the spatiotemporal distribution matrix of future reachable anchor points, generating control instructions based on the anchor point activation priorities in the matrix, and feeding them back to the dynamic anchor point collaborative construction module is as follows: The spatiotemporal reachability probability distribution map is quantized into a matrix with the dimensions of time-space-anchor identifier. The spatiotemporal nodes in the matrix are sorted by their probability values to generate anchor activation priority instructions, giving priority to waking up anchors with high probability of being reached. The priority instructions are fed back to the dynamic anchor point collaborative construction module through the port's wireless private network, triggering the anchor point wake-up or sleep operation, and monitoring the anchor point activation effect in real time. If the vehicle fails to reach the anchor point as predicted, the priority instructions are dynamically adjusted and the matrix is regenerated.

8. A method for positioning a fleet inside or outside a port, applied to a system for positioning a fleet inside or outside a port according to any one of claims 1 to 7, characterized in that: The following steps are involved: S1. Real-time acquisition of dynamic location data of mobile equipment in the port, dynamic calculation of anchor point signal coverage based on the signal propagation attenuation characteristics in metallic environments, and generation of anchor point activation instructions and anchor point coverage blind spot information based on cluster analysis of historical vehicle trajectories; S2. Receive anchor point coverage blind spot information, perform local grayscale entropy segmentation on the ground image captured by the vehicle-mounted camera, select high-resolution texture areas, and eliminate dynamic interference through cross-frame Bayesian probabilistic association to generate pose estimation data that is temporally and spatially aligned with the anchor point signal. S3. Pose estimation data and vehicle inertial navigation data are combined with inter-vehicle communication topology to construct an error propagation chain model, and the accumulated positioning error is corrected through a multi-vehicle confidence weight fusion strategy. S4. Analyze the operation plan of the mobile equipment in the port scheduling system, build a spatiotemporal reachability prediction model based on the motion state of the corrected vehicle posture, predict the spatiotemporal distribution matrix of future reachable anchor points, and generate control instructions based on the anchor point activation priority in the matrix and feed them back to step S1.

Images