Custom map large screen generation method and system, product and medium

Abstract

The invention discloses a custom map large screen generation method and system, a product and a medium. The method comprises the following steps: when position information of a target object is interrupted in a signal blind area, acquiring macroscopic traffic data of the area to determine a dynamic deduction speed; continuously calculating the deduction position of the object according to the speed and the blind area path, updating the icon display on the large screen of the map by using the position, and switching the icon into a special blind area rendering mode; if the object is still not driven out beyond the preset time, triggering an overtime alarm; once the real-time data is recovered, the icon is switched back to the conventional mode and is instantly calibrated to the latest real position. By implementing the technical scheme provided by the invention, the accuracy of monitoring the motion trail when the target object on the large screen enters the signal blind area is improved.

Description

Technical Field

[0001] This application relates to the field of indicating the location of vehicles on maps, and more particularly to a method, system, product, and medium for generating custom map screens. Background Technology

[0002] Currently, large-scale digital monitoring platforms are becoming a core component of modern management systems. With the rapid development of technologies such as the Internet of Things and mobile communications, the real-time collection and visualization of location data for moving targets such as vehicles and personnel are being increasingly widely applied in fields such as logistics, public transportation, and emergency command.

[0003] In related technologies, a combination of trajectory estimation and map matching is commonly used to address data interruption issues in signal blind spots. This method, after signal interruption, uses the target object's last valid kinematic state information, such as position, velocity, and direction, to mathematically estimate its subsequent trajectory. Simultaneously, a map matching algorithm constrains the estimated trajectory points onto a reasonable road network on the map to correct accumulated errors and ensure the estimated path conforms to geographical logic.

[0004] However, the relevant technical calculation process is highly dependent on the static kinematic state at the moment of interruption. When a target enters a long blind zone, such as a river-crossing tunnel or a mountain road section, the internal traffic environment may change dynamically and unpredictably. Once the actual traffic flow speed inside the blind zone slows down significantly or even stops due to congestion, the movement simulation will still be based on the higher speed at which it entered. This will cause the position displayed on the screen to deviate significantly from the target's true position, reducing the accuracy of judging the target's movement trajectory and making it difficult to truly reflect the target's movement state, thus misleading command and dispatch. Summary of the Invention

[0005] This application provides a method, system, product, and medium for generating custom map screens, which can improve the accuracy of the movement trajectory of target objects entering signal blind spots on monitoring screens.

[0006] The first aspect of this application provides a method for generating a custom map large screen, the method comprising: The system acquires real-time data of the map component used to generate the large screen, as well as one or more target objects bound to the map component. When the real-time location of a target object is interrupted and the last acquired real-time location is within a preset signal blind zone geofence, the system records the last effective speed, interruption location, and interruption time of the corresponding blind zone target object before the information interruption. Based on macro traffic data, the system determines the projected speed and, combined with the signal blind zone path, continuously calculates the projected location and estimated exit time of the blind zone target object within the signal blind zone. The system updates the display of the blind zone target object's icon on the map component using the projected location, and simultaneously changes the visual style of the blind zone icon corresponding to the blind zone target object to the preset blind zone rendering mode. If the real-time data of the blind zone target object is not recovered after the estimated exit time, a timeout alarm message is generated and displayed on the large screen. When the real-time data of the blind zone target object is detected to be recovered, the rendering method of the blind zone icon is switched back to the normal mode and calibrated to the latest real-time location.

[0007] In the above embodiments, to address the shortcomings of relying solely on the static speed of the target before its interruption for prediction, external, dynamic macroscopic traffic data is introduced as the core basis for prediction. This macroscopic traffic data can reflect the real traffic flow status within the signal blind zone (such as inside a tunnel) in real time. Therefore, when congestion occurs within the blind zone and the actual vehicle speed decreases, this change can be detected, and the predicted speed can be adjusted accordingly, thus breaking the rigid dependence on the initial speed of the target when entering the blind zone. This dynamic speed calibration mechanism based on real-time regional road conditions suppresses the continuous deviation between the predicted and actual positions caused by changes in the environment within the blind zone, ultimately achieving high-fidelity dynamic prediction of the target trajectory within the blind zone, improving the accuracy and reliability of the prediction.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the projected speed is determined based on macroscopic traffic data, and combined with the signal blind zone path, the projected position and estimated exit time of the target object within the signal blind zone are continuously calculated, specifically including: The signal blind zone path is pre-divided into a predetermined number of ordered path sub-segments in the map data; a historical passage queue consisting of the most recently maintained predetermined number of similar target objects that have successfully exited the signal blind zone is maintained in real time, and the actual total passage time of each similar target object in the queue is recorded; based on the actual total passage time, a baseline extrapolated time is calculated using a weighted average algorithm, and combined with macro traffic data, the final extrapolated total time is determined; the final extrapolated total time is allocated to each path sub-segment according to a predetermined speed distribution model, and an independent extrapolated sub-speed is generated for each path sub-segment; when a target object enters the signal blind zone, the extrapolated position is calculated segment by segment according to the corresponding extrapolated sub-speed until the travel trajectory covers all path sub-segments, and the estimated exit time is determined based on the final extrapolated total time.

[0009] In the above embodiments, by constructing historical traffic queues and combining them with macroscopic traffic data, a projected total travel time is generated that is both historically relevant and consistent with current realities, ensuring the macroscopic accuracy of the time prediction. By segmenting the path and applying a speed distribution model, a macroscopic time prediction is transformed into a microscopic dynamic speed simulation that closely resembles real driving habits. Ultimately, this improves the rationality and credibility of the projected trajectory during the process.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, a baseline estimated travel time is calculated using a weighted average algorithm based on the actual total travel time, and the final estimated total travel time is determined by combining macroscopic traffic data, specifically including: At the entrance to the signal blind zone, by monitoring the time interval of target objects entering, the traffic flow clusters formed by continuously entering vehicles are identified in real time, and multiple target objects within the same traffic flow cluster are defined as a dynamic queue of traffic flow clusters with internal order. When a target object in the blind zone enters the signal blind zone, its order position in the dynamic queue of its traffic flow cluster is determined. When executing the weighted average algorithm, an order position weight factor is introduced to assign the first calculation weight to the actual total passage time of historical target objects that have the same order position as the target object in the blind zone in the historical passage queue and in their corresponding traffic flow clusters. Based on the weighted average algorithm with the introduced order position weight factor, the final estimated total passage time is calculated.

[0011] In the above embodiments, by constructing a dynamic queue of traffic flow clusters, we can delve into the micro-dynamic relationships within the traffic flow from a macro-statistical perspective, identify the order position of the target object in the queue, and introduce an order position weight factor to give higher computational weight to samples with the same queue order in historical data. This enables the inference model to accurately match the target's current following state with the most similar traffic scenario in history, thereby improving the accuracy and reliability of the total travel time prediction.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, the projected speed is determined based on macroscopic traffic data, and combined with the signal blind zone path, the projected position and estimated exit time of the target object within the signal blind zone are continuously calculated, specifically including: Before a target object enters the signal blind zone, a normal behavior state vector is collected. Based on the normal behavior state vector, a path selection confidence level is calculated for each possible branch path in the signal blind zone using a preset path selection probability model. Based on the path selection confidence level and macro traffic data, parallel simulations are performed on the map component: along the main path with the highest path selection confidence value, the main simulation position is continuously calculated and rendered, and the corresponding main estimated departure time is determined; along the secondary paths with the corresponding confidence levels of all other paths besides the main path, the secondary simulation positions are continuously calculated and rendered, and the secondary estimated departure time is determined; all main simulation positions, main estimated departure times, secondary simulation positions, and secondary estimated departure times are presented side-by-side in a visual representation.

[0013] In the above embodiments, multiple possibilities are quantified by introducing path selection confidence, and parallel inference is performed based on this. By visualizing the primary and secondary inference paths side by side, a global decision-making view containing all reasonable possibilities is provided to the commanders, thereby improving the command and decision-making capabilities in complex scenarios.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, after selecting secondary paths corresponding to confidence levels along all paths other than the main path, continuously calculating and rendering the secondary projected positions, and determining the secondary estimated departure time, the method further includes: When a secondary or primary path contains a preset internal verification point, it becomes a verification path containing the internal verification point. The estimated arrival time of the target object in the blind zone on the verification path is calculated. The data interface of the internal verification point is monitored, and within a preset monitoring time window before and after the estimated arrival time, the passability verification data of the target object in the blind zone is checked. If so, the parallel simulation and visualization of all other paths except the verification path are terminated and cleared. The simulation of the verification path is calibrated and continued based on the internal verification point. If not, the confidence of the path selection of the verification path is dynamically reduced to a preset extremely low value, and the confidence of all other paths except the verification path is increased according to an average distribution ratio.

[0015] In the above embodiments, by introducing internal verification points as trusted data relays within the blind zone and combining this with real-time monitoring of passability verification data, multiple parallel paths can be dynamically verified or disproven during the simulation process. This achieves convergence and focusing of the simulation paths in an uncertain environment, ensuring that the final trajectory presented to the decision-maker possesses both high probability and high credibility.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, before switching the rendering method of the blind spot icon back to normal mode and calibrating it to the latest real-time position when real-time data recovery of the blind spot target object is detected, the method further includes: When the geographical deviation exceeds a preset deviation threshold, the trajectory deviation point that is geographically closest to the actual restored location is located in the historical projection trajectory of the target object in the blind spot that has been rendered on the map component. Based on the underlying road network data of the map component, a path planning is performed to connect the two points, starting from the trajectory deviation point and ending at the actual restored location, to generate a corrected path segment. The portion of the historical projection trajectory from the trajectory deviation point to its end is replaced with the corrected path segment, and then the blind spot icon is calibrated to the latest real-time location.

[0017] In the above embodiment, by tracing back the historical trajectory, the deviation point is located, thus translating an isolated position error into a continuous path deviation problem. Based on this, path planning is performed using underlying road network data to generate corrected path segments with reasonable path rationality and then perform replacement. Ultimately, not only is the current position calibrated, but the historical trajectory is also logically and interpretably reconstructed, ensuring the continuity and reliability of the trajectory throughout the blind spot passage process, achieving a qualitative leap from error correction to trajectory restoration.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, after generating and displaying a timeout alarm message on the large screen when real-time data recovery of the target object in the blind spot has not been obtained after the expected departure time has exceeded, the method further includes: Centered on the preset exit point of the signal blind zone, when a neighboring signal blind zone is found within a preset geographical radius, the visual style of the timeout alarm information of the target object in the signal blind zone is changed to an inactive mode indicating that the simulation has been completed but the status is unknown; at the entrance of the neighboring signal blind zone, a highly uncertain speculation icon is generated, along with new blind zone alarm information.

[0019] In the above embodiment, two independent blind zone events were logically linked by searching for adjacent signal blind zones. Based on this, the termination of the previous deduction was marked by converting the timeout alarm of the previous blind zone to an inactive mode, avoiding confusion of states; simultaneously, the generation of new blind zone alarm information bridged the uncertainty of the next stage. Ultimately, two isolated, potentially misjudgment-causing disconnection alarms were reconstructed into a continuous and predictable blind zone transition event, avoiding decision-making errors caused by information fragmentation.

[0020] Secondly, embodiments of this application provide a custom map large screen generation system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, which includes computer instructions, and the one or more processors call the computer instructions to cause the custom map large screen generation system to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a custom map large screen generation system, cause the custom map large screen generation system to execute the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a custom map large screen generation system, cause the custom map large screen generation system to perform the method described in the first aspect and any possible implementation thereof.

[0023] Understandably, the custom map large-screen generation system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the custom map large-screen generation method provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. This application addresses the shortcomings of related technologies that rely solely on the static speed of the target before its interruption for prediction. It introduces external, dynamic macroscopic traffic data as the core basis for prediction. This macroscopic traffic data can reflect the real-time traffic flow status within signal blind zones (such as tunnels). Therefore, when congestion occurs within the blind zone and the actual vehicle speed decreases, this change can be detected, and the predicted speed can be adjusted accordingly, thus breaking the rigid dependence on the initial speed of the target entering the blind zone. This dynamic speed calibration mechanism based on real-time regional traffic conditions suppresses the continuous deviation between the predicted and actual positions caused by changes in the blind zone environment. Ultimately, it achieves high-fidelity dynamic prediction of the target trajectory within the blind zone, improving the accuracy and reliability of the prediction.

[0025] 2. This application constructs historical traffic queues and combines them with macroscopic traffic data to generate a projected total travel time that is both historically relevant and consistent with current realities, ensuring the macroscopic accuracy of the time prediction. By segmenting the path and applying a speed distribution model, a macroscopic time prediction is transformed into a microscopic dynamic speed simulation that closely resembles real driving habits. Ultimately, this improves the rationality and credibility of the projected trajectory during the process.

[0026] 3. This application introduces internal verification points as trusted data relays within the blind zone and combines this with real-time monitoring of passability verification data, enabling dynamic verification or falsification of multiple parallel paths during the simulation process. This achieves convergence and focusing of the simulation paths in an uncertain environment, ensuring that the final trajectory presented to decision-makers possesses both high probability and high credibility. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating a custom map large screen generation method in an embodiment of this application; Figure 2 This is another flowchart illustrating the custom map large screen generation method in this application embodiment; Figure 3 This is an exemplary hardware structure diagram of a custom map large screen generation system in the embodiments of this application. Detailed Implementation

[0028] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0030] In related technologies, to address data interruption caused by signal blind spots, a common approach is to extrapolate the trajectory based on the target's own kinematic state. After the target's signal is interrupted, information such as its position, velocity, and direction at the last moment is extracted as a static baseline for extrapolation. This is then combined with map matching technology to perform linear extrapolation of the trajectory over a road network. The underlying logic of this method is the assumption that the target will maintain its pre-entry motion trend within the blind spot. However, in real-world scenarios, especially in long tunnels or complex mountainous road sections, internal traffic conditions can change drastically due to sudden congestion, accidents, or other factors. If the actual traffic speed within the blind spot differs significantly from the target's entry speed, the extrapolation algorithm will still calculate based on outdated and inaccurate initial speeds. This leads to a continuously widening deviation between the extrapolated trajectory and the actual location, ultimately outputting severely distorted location information and misleading command and dispatch situational awareness and decision-making.

[0031] In this embodiment, external, dynamic macroscopic traffic data is introduced as the core decision-making basis for speed estimation. Instead of assuming a constant target state, real-time traffic flow status within signal blind spots is obtained by periodically requesting third-party traffic data services. Therefore, even if the target encounters congestion after entering the blind spot, this solution can immediately obtain the objective fact of decreased road network speed and dynamically adjust the estimated speed to keep pace with the actual traffic flow in the area. This paradigm shift from relying on individual history to referencing the reality of the environment breaks the rigid dependence on initial speed, ensuring that the estimation process can dynamically adapt to real changes within the blind spot, achieving high-fidelity dynamic simulation of the target trajectory within the blind spot, and improving the accuracy and reliability of location estimation.

[0032] Figure 1 This is a flowchart illustrating the custom map large-screen generation method used in this application embodiment, including the following steps: S101. Obtain the real-time data of the map component used to generate the large screen and one or more target objects bound to the map component.

[0033] The map component refers to a programmable module on the front end of a software application that is responsible for rendering geospatial data and providing interactive functions. It originates from a human-defined setting based on project technology selection and business needs. For example, depending on factors such as whether the project operates domestically, requirements for 3D effects, or development costs, the Gaode Map JSAPI 2.0 might be pre-set as the map rendering engine. The target object refers to a physical or logical entity that has been given a unique identifier and whose status (especially its geographical location) is continuously monitored by the system. This is the initial configuration or backend management setting of the business system. For example, in a logistics fleet monitoring system, the administrator pre-associates the list of vehicle IDs to be monitored with the large screen in the system backend. Real-time data refers to the structured data stream continuously reported by the terminal devices mounted on the target object at a preset second-level or sub-second-level frequency. Rendering represents the process of converting abstract data coordinates into visible pixels on the screen through graphical calculations, while the icon is a specific visual symbol representing the target object generated on the map layer during this process, such as a vector arrow that automatically rotates according to the driving direction. The target object's data includes its real-time location, and its icon is rendered on the map component based on the real-time location.

[0034] Specifically, the backend data access gateway maintains uninterrupted communication with hundreds or thousands of target terminal devices via TCP long connections or MQTT protocols. Once it receives a location data packet reported by any terminal, the backend service parses, cleans, and performs coordinate system transformation, stores it in a high-speed cache, and immediately pushes the formatted JSON data to the frontend map component via full-duplex communication channels such as WebSocket. When the frontend JavaScript logic detects new data, it first checks if the target object's icon already exists on the map. If it doesn't, it instantiates a new map marker or layer object and adds it to the map; if it already exists, it calls the object's setPosition or a similar method to smoothly update the screen position. This receive-push-update loop macroscopically creates a dynamic monitoring screen showing the smooth movement of numerous target icons on the large display.

[0035] S102. When the real-time location information of the target object is interrupted and the last obtained real-time location is within the preset signal blind zone geofence, record the last effective speed, interruption location and interruption time of the target object in the corresponding blind zone before the information interruption.

[0036] Among them, "information interruption" refers to a system-level state judgment. When no data is received from a specific target object within several consecutive preset communication heartbeat cycles, an interruption is determined. "Signal blind zone geofence" refers to a closed geographical area pre-defined in the form of a vector polygon using a set of ordered coordinate points in a geographic information system. Logically, it represents a specific area in the physical world where cellular network or satellite signal coverage is weak or absent, such as river tunnels or underground loops. "Blind zone target object" refers to a target object whose last known location is determined to fall within any preset geofence after spatial geometric calculations when an information interruption occurs. "Last effective velocity," "interruption location," and "interruption time" refer to the instantaneous velocity value, latitude and longitude coordinates, and UTC timestamp recorded in the last telemetry message successfully received and verified by the server reported by the target object, respectively.

[0037] Specifically, a status timer is maintained for each target object in transit. Under normal circumstances, the last communication timestamp of the object is refreshed every time data is received. The background daemon periodically scans the status of all objects. Once it finds that the difference between the current system time and the last communication timestamp of an object exceeds a preset time threshold (based on experience with device reporting frequency and network fluctuation tolerance; for example, if a device is configured to report data every 10 seconds, a preset threshold of 30 seconds (i.e., 3 reporting cycles) can be used to filter out occasional network delays), an interruption handling process is triggered. This process first locks the last record of the object in the database or cache and extracts its latitude and longitude coordinates. Then, it calls the ST_Contains function of a spatial database (such as PostGIS) or a geometric calculation library (such as JTS) to traverse all preset signal blind zone geofences and determine whether the coordinate point is contained within any of the fences. If the result is true, it confirms that it is a signal interruption event within the blind zone, rather than a normal device offline event. At this point, the internal state of the object is marked as "BLIND_ACTIVE", and the three key parameters of speed, position and time recorded at the end are packaged into an initial simulation task and stored in a dedicated, highly available task queue (such as Redis List or RabbitMQ).

[0038] S103. Determine the projected speed based on macro traffic data, and combine the signal blind zone path to continuously calculate the projected location and estimated exit time of the target object in the signal blind zone.

[0039] Macro-level traffic data refers to quantitative indicators provided by third-party data service providers that describe the overall traffic flow situation within a specific geographical area. Examples include regional average speed, road congestion delay index, or real-time traffic condition levels (e.g., smooth, slow, congested) calculated using floating car big data. This data is obtained by repeatedly requesting and retrieving data from a third-party traffic data service interface at a preset time period, based on the geographical location information of signal blind spots. The preset time period is a manually set value that balances data real-time performance with interface call costs / frequency limitations, taking into account factors such as the relative changes in macro-level traffic conditions. The API is flat and has a QPS (queries per second) limit, defaulting to requesting data from a third-party interface once every 60 seconds; the projected speed is a virtual speed scalar dynamically calculated to simulate driving in blind spots, which is a function of macro traffic conditions and individual historical behavior data; the signal blind spot path refers to the vector geometric representation of one or more deterministic driving paths that are predefined in the digital road network model and connect the entrance and exit of the blind spot; the estimated exit time is a future point in time calculated using kinematic formulas based on the projected speed and the total length of the signal blind spot path, representing a prediction of when the target object will regain signal.

[0040] Specifically, after retrieving a simulation task (containing initial speed, location, and time) from the task queue, a simulation session is initiated. This session first periodically (e.g., every 60 seconds) sends an HTTP request to a third-party traffic data service interface (such as Baidu Maps' "Regional Traffic Conditions" API) based on the blind spot identifier in the task. The request parameter is the geographical coordinate range of the blind spot. The macro-level traffic data returned by the API (e.g., {"status": "congestion", "avg_speed": 15}) is used as a dynamic correction factor. The process of determining the simulated speed is a multi-factor fusion calculation: based on the instantaneous speed of the target object before entering the blind spot, a weighted average is calculated based on the obtained macro-level average speed, or by referring to historical data on the time taken by other similar targets to pass through the blind spot, and then adjusting for current macro-level traffic conditions, resulting in a highly contextualized simulated speed. Once the simulated speed is determined, a timer is started. Within each time step (Δt), starting from the previous projection point, a new projection position is generated by iteratively calculating the new position = old position + projection speed × Δt, along the preset blind zone path vector direction. Simultaneously, the estimated total time is calculated based on the total path mileage and the current projection speed. Adding the interruption time yields the final estimated departure time.

[0041] In some embodiments, the determination of dynamic extrapolation speed and the continuous generation of trajectory can be achieved in various ways: Optionally, a machine learning method based on historical data regression can be adopted. The specific process is as follows: First, the system collects and stores in the background the actual passage time of all target objects that successfully pass through each blind zone, as well as the corresponding macro traffic index, entry time, weather and other features during the passage period; Second, a regression model (such as gradient boosting decision tree GBDT) is trained using these historical data. This model can predict the most likely total passage time based on the current input features; Finally, when a new target enters, the model is called to make a prediction. After obtaining the total passage time, the average extrapolation speed is calculated, and the position extrapolation is performed based on this, whether at a constant speed or a variable speed.

[0042] Understandably, more complex micro-traffic flow simulation models can be used to achieve more refined simulations of vehicle behavior in blind spots, which is not limited here.

[0043] In some embodiments, a refined extrapolation method combining micro-traffic flow characteristics and macro-speed distribution models can be used to achieve a prediction effect that is closer to actual driving behavior.

[0044] Specifically, firstly, to model the speed-changing behavior within blind spots, each signal blind spot path is pre-divided into several ordered path sub-segments in the underlying road network data based on geometric features (such as curvature and slope) or historical accident data. A long-distance process with potentially constantly changing speed is discretized into multiple short-distance sub-processes that can be assumed to be at a constant speed or uniform acceleration. Simultaneously, a fixed-capacity, first-in-first-out historical passage queue is maintained in real-time for each signal blind spot. This queue specifically stores telemetry records of the most recently preset queue value (an empirical parameter set to balance statistical stability and data timeliness, aiming to ensure a sufficient sample size to smooth out the randomness of individual driving behavior while quickly responding to recent dynamic changes in traffic conditions; for example, set through simulation testing or experience to the most recent 20) successfully exiting the blind spot, and of the same type as the current target object (such as a small passenger car). The actual total travel time for each of these vehicles through the entire blind spot is stored, calculated, and recorded. This queue forms a dynamically updated, highly correlated historical experience database, which can better reflect the real traffic efficiency of similar vehicles in the recent period compared to the global statistical average.

[0045] When a target object enters the blind spot, the system first identifies whether a traffic cluster has formed by monitoring the time interval between its entry and the vehicle ahead's entry. For example, if multiple vehicles enter at intervals shorter than a preset traffic cluster time length (based on statistical analysis of historical traffic data at the entrance, or referencing the safe headway in traffic engineering theory, such as the industry-recognized two-second rule as a benchmark), they are considered to belong to the same traffic cluster. A temporary dynamic queue is created for this cluster, and each vehicle is assigned a unique position (e.g., first, second, third) based on its entry order. In enclosed environments such as tunnels, the driving behavior of vehicles within the traffic flow is strongly constrained by the vehicle ahead, and the travel time is highly correlated with their position within the traffic flow.

[0046] Therefore, a special weighted average algorithm incorporating the order position weight factor is used when calculating the estimated travel time. Specifically, when calculating the estimated travel time for a blind spot target object with order position N, every historical object in the historical traffic queue is traversed. If a historical object is also at position N in its current traffic cluster, its actual total travel time is assigned the highest weight; while the travel times of other historical objects (including vehicles outside the cluster or vehicles at different positions within the cluster) are assigned lower, conventional weights. The weighted average calculated in this way is the baseline estimated travel time, maximizing the reference to historical samples most similar to the current target object's traffic micro-environment, thus improving the relevance and accuracy of the baseline value.

[0047] Finally, this baseline extrapolation time is fused with the macroscopic traffic data obtained in S103 (e.g., by performing secondary weighting or as a correction factor) to obtain the final total extrapolation time.

[0048] By leveraging long-term data mining, a speed distribution model is established for each path sub-segment of the blind zone. This model is essentially a normalized coefficient or function describing the trend of speed changes across different segments. For example, the model might show that the vehicle accelerates in the first 10% segment after the entrance, decelerates to its minimum in the middle curve segment, and accelerates again on the straight section before the exit. The final total simulation time calculated in the second stage is allocated to each path sub-segment according to the proportion defined by this speed distribution model, thus calculating an independent and distinct simulation sub-speed for each segment. Finally, the simulation process is no longer a constant speed throughout; instead, as the target object enters the blind zone, the position is calculated segmentally and stepwise according to the simulation sub-speed corresponding to each sub-segment until the trajectory covers all sub-segments. The estimated exit time is directly determined by adding the interruption time to the final total simulation time. The generated simulation trajectory is no longer a monotonous straight line or curve, but a dynamic trajectory that includes acceleration and deceleration processes and is more consistent with real driving habits.

[0049] The aforementioned technical steps, by identifying traffic flow clusters and introducing a sequence position weighting factor, prioritize the calculation of estimated time based on historical samples with the most similar traffic microstructure, thus improving the accuracy of time prediction. Furthermore, by utilizing a speed distribution model, the total time is transformed into a non-uniform segmented trajectory, making the spatial location prediction more closely resemble real driving behavior throughout the process. Therefore, this dual refinement in both time and space dimensions jointly improves the consistency between the final estimated trajectory and predicted departure time and the actual future situation.

[0050] S104. Use the inferred location to update the display of the icon of the blind spot target object on the map component, and at the same time change the visual style of the blind spot icon corresponding to the blind spot target object to the preset blind spot rendering mode.

[0051] Among them, the projected location is a series of simulated geographic coordinates continuously output by the iterative calculation process described in step S103, which together constitute the predicted trajectory of the target object during the signal interruption; updating the icon of the target object in the blind zone refers to dynamically and continuously migrating the position of the graphic marker representing the object on the front-end map from the last real coordinate point to the projected location calculated in real time; visual style refers to the set of all configurable graphic attributes that constitute the appearance of the icon, covering fill color, border color, transparency, size, shape, and even the bound animation effects; preset blind zone rendering mode refers to a set of pre-designed and coded specific visual style schemes that are different from the normal state, which clearly convey to the user through significant visual differences that the current position information of the icon is based on algorithmic speculation rather than real-time data.

[0052] Specifically, each time a new calculated location point is determined, it is encapsulated into a data packet along with the corresponding target object ID and pushed to the front-end large-screen application in real time via a WebSocket connection. Upon receiving such a message designated as a calculation update, the front-end event listener executes a rendering function. This function first checks the local state of the target object. If it's the first time the object's calculation data has been received (i.e., it has just entered the blind zone), a one-time style switching operation is performed: the icon object is retrieved, and its rendering properties are batch-modified to the preset blind zone rendering mode. For example, a solid blue vehicle icon is changed to a semi-transparent gray icon with a dashed outline, and a regular scaling animation effect may be added. Subsequently, regardless of whether it's the first time, the function performs a location update operation, that is, it calls a method provided by the map engine to set the icon's geographic coordinates to the latest calculated location carried in the message. This calculation-push-rendering closed loop repeats continuously at a high frequency (e.g., once per second), so that the user sees a uniquely styled icon that is still smoothly moving on the screen, thus maintaining visual continuity of tracking while clearly conveying the uncertain state of the data.

[0053] S105. If real-time data recovery of the target object in the blind spot is not obtained after the expected departure time, a timeout alarm message is generated and displayed on the large screen.

[0054] Among them, exceeding the expected departure time means that when the current clock time has exceeded the predicted recovery signal time point calculated for a certain blind zone target object in step S103, and no real-time data of the object has been received, the condition is met; timeout alarm information refers to a high-priority notification generated and presented to the user after the above timeout condition is triggered. It can take the form of a new text record in the alarm list, a significant status change of the icon on the map (such as turning red and flashing), or a modal pop-up window.

[0055] Specifically, after the estimated departure time of a target object is calculated in step S103, a one-time, timed scheduled task is registered in the background. The scheduled execution time of the task is the estimated departure time. After being scheduled, this task enters a waiting state. If step S106 is triggered during the waiting period, i.e., the recovery of the object's data is detected, a cancellation command is sent to the task scheduler to prematurely cancel the waiting, timed-out task. Conversely, if the scheduled execution time has elapsed and the task has not been canceled, the scheduler will execute the task's predefined logic. The core of this logic is to create a structured alarm event object, which records details such as the target ID, the blind zone it belongs to, the signal interruption time, the estimated departure time, and the current timeout time. This alarm event is then pushed to the system's unified alarm center. The alarm center, according to preset rules, pushes the information to the front-end dashboard via WebSocket and may simultaneously notify relevant administrators via SMS or email. Upon receiving the alarm push, the front-end will immediately perform a corresponding UI update, presenting the timeout status to the user in a prominent manner.

[0056] In some embodiments, when a target object fails to leave within the specified time, in order to avoid false alarms caused by continuous or adjacent signal blind spots in the road network, intelligent judgment can be further performed by combining the road network topology relationship, so as to distinguish the timeout state and infer the continuity of the target trajectory.

[0057] First, the triggering condition is defined by S105: the current system time has exceeded the previously calculated estimated departure time, and the monitoring system has not yet received any real-time telemetry data of the target object in the blind zone. At this time, a timeout alarm is generated.

[0058] Secondly, this timeout alarm will not be immediately pushed to monitoring personnel with the highest priority. Instead, an automatic background retrieval and analysis program will be executed. Specifically, taking the geographic coordinates of the preset exit point of the signal blind zone as the center, within a preset geographic radius (this radius is an empirical parameter, set to balance retrieval efficiency and association accuracy, mainly determined based on the maximum normal connection distance between typical urban road intersections, overpasses, or adjacent tunnel entrances, for example, it can be set to 50 meters to ensure that it can cover the scenario where the vehicle immediately enters the next nearby blind zone after exiting, while avoiding incorrect association with unrelated distant blind zones due to an excessively large radius), the GIS road network database will be queried to see if there are other predefined neighboring signal blind zones (referring to another predefined signal blind zone in the GIS road network topology that has a direct, legal, and passable road connection to the preset exit point of the current signal blind zone, and the path distance between the two is extremely short, so that the target object may not be able to effectively re-establish and report stable telemetry data during normal driving) entrances).

[0059] This search has two possible results: Scenario 1: No entrance to any nearby signal blind zone was found, meaning that from a topological perspective, the target vehicle should have entered an area with good signal coverage after leaving the current blind zone. In this case, failure to restore the signal within the timeout period is a highly abnormal event, most likely caused by equipment failure, severe congestion, or a traffic accident. Therefore, the original processing logic of S105 is maintained, and the timeout alarm information is displayed on the large screen in a standard or highlighted form (such as a flashing red icon) to prompt monitoring personnel to pay immediate attention.

[0060] Scenario 2: Successfully detected the entrance to one or more adjacent signal blind spots. This situation is very common in complex urban road networks (such as continuous tunnels, large underground ring roads, multi-level overpasses, etc.). Its occurrence provides a reasonable, non-urgent explanation for the timeout: the target object is very likely to have entered the next signal blind spot directly after leaving the current blind spot without passing through any signalized intermediate areas.

[0061] The original alarm is downgraded and its status changed. The visual style of the timeout alarm message for the target object in the first (i.e., the timed-out) signal blind zone is changed from a standard, abnormally active state (such as flashing red) to a special inactive mode. This mode (e.g., turning the icon gray, stopping the flashing, and adding a checkmark indicating "path segment completed") is intended to convey a clear message to monitoring personnel: the simulation task for this blind zone has been completed as planned, and the reason why the target did not appear on time is likely not due to an anomaly, but rather because it has entered the next path segment.

[0062] Simultaneously, on the large-screen map, a highly uncertain speculative icon is generated at the entrance of the detected adjacent signal blind zone. This icon is visually distinct from the actual target (e.g., using semi-transparency, a dashed outline, or a question mark superimposed on the icon) and includes a clear new blind zone warning message (such as the text "Target may have entered XX tunnel"). The value of this operation lies in continuing the monitoring chain, guiding the monitoring personnel's attention to the new location where the target is most likely to appear, and clearly indicating that the current status is based on logical inference rather than factual confirmation, thus achieving continuous tracking of the target's trajectory and prediction of future states.

[0063] The above technical steps, by introducing road network topology analysis, perform in-depth semantic parsing of the single alarm event of timeout. The derivation logic is as follows: timeout is further distinguished into two scenarios—high-probability anomalies and high-probability continuous crossings—by searching for nearby blind spots. This distinction allows the alarm system to filter out benign timeouts caused by the road network structure itself, thereby reducing the false alarm rate. This allows monitoring personnel to focus their limited energy on truly abnormal events requiring manual intervention, improving the overall operational efficiency of the monitoring system.

[0064] S106. When real-time data recovery of the target object in the blind zone is detected, switch the rendering method of the blind zone icon back to normal mode and calibrate it to the latest real-time position.

[0065] Among them, real-time data recovery of the detected blind zone target object refers to the server's data access layer receiving a legitimate and valid location data packet from a terminal device whose ID is marked as being in blind zone simulation or in a timeout state after a period of communication silence; normal mode refers to the default rendering style of the target icon representing the true and reliable data under normal signal conditions; calibration to the latest real-time position is to move the icon on the map from its last simulation position or timeout position to the first real geographic coordinate point reported after data recovery.

[0066] Specifically, the real-time data receiving service parses all uplink data packets and extracts unique device identifiers. Internally, the service queries a state table (potentially based on Redis or memory) that maintains the current state of all devices. When a data packet's device identifier shows a current state of "BLIND_ACTIVE" or "BLIND_TIMEOUT," it determines that the target has left the blind zone, and data recovery is complete. This determination immediately triggers synchronization actions: on the backend, all calculations for the target are immediately stopped, and related scheduled tasks (such as timeout alarm tasks) are cleaned up. Simultaneously, a state change instruction is encapsulated, explicitly indicating the target ID, the new state (e.g., "ONLINE"), and the first location coordinates after data recovery. This instruction is sent to the frontend via WebSocket with high priority. Upon receiving the instruction, the frontend application executes a recovery function. This function resets the icon's style object to the default normal mode configuration and calls the map engine's setPosition or setCoordinates method to directly set the icon's coordinates to the latest real-time location specified in the instruction.

[0067] In some embodiments, state synchronization and location calibration after data recovery can be achieved in several ways: Optionally, a centralized state machine management mechanism can be used, with the following process: First, the backend maintains a finite state machine for each device, defining states such as Online, Blind, and Timeout, as well as the transition rules between them; second, a data recovery event triggers a DATA_RECEIVED input, driving the state machine to transition from the Blind or Timeout state back to the Online state; finally, each state transition is accompanied by triggering one or more actions, including pushing instructions for state changes and the latest location to the frontend. This approach makes the entire state management logic very clear and robust.

[0068] Understandably, a path planning algorithm can also be used to draw a corrected path from the simulation endpoint to the actual location before calibration, in order to explain the positional deviation more visually. This is not a limitation here.

[0069] In the above embodiments, to address the shortcomings of relying solely on the static speed of the target before its interruption in related technologies, external, dynamic macroscopic traffic data is introduced as the core basis for the extrapolation. This macroscopic traffic data can reflect the real traffic flow status within the signal blind zone (such as inside a tunnel) in real time. Therefore, when congestion occurs within the blind zone and the actual vehicle speed decreases, this change can be detected, and the extrapolated speed can be adjusted accordingly, thus breaking the rigid dependence on the initial speed of the target when entering the blind zone. This dynamic speed calibration mechanism based on real-time regional road conditions suppresses the continuous deviation between the extrapolated position and the actual position caused by changes in the environment within the blind zone, ultimately achieving high-fidelity dynamic extrapolation of the target trajectory within the blind zone, improving the accuracy and reliability of the prediction.

[0070] In some other embodiments of this application, when the initial confidence levels of multiple parallel inference paths are highly similar, the prediction may become stuck due to a lack of decision-making basis, failing to converge quickly. The custom map large-screen generation method provided in this application breaks the deadlock and improves convergence efficiency by actively falsifying key nodes and eliminating erroneous paths.

[0071] like Figure 2 The diagram shown is another flowchart illustrating the custom map large-screen generation method provided in this application embodiment, including the following steps: S201. Obtain the real-time data of the map component used to generate the large screen and one or more target objects bound to the map component.

[0072] S202. When the real-time location information of the target object is interrupted and the last obtained real-time location is within the preset signal blind zone geofence, record the last effective speed, interruption location and interruption time of the target object in the corresponding blind zone before the information interruption.

[0073] Steps S201-S202 and Figure 1 Steps S101-S102 in the illustrated embodiment are similar and can be found in the descriptions of steps S101-S102, which will not be repeated here.

[0074] S203. Before the target object enters the signal blind zone, collect the normal behavior state vector.

[0075] The blind spot target object refers to a specific traffic participant that the current monitoring system is tracking and is about to enter a predefined area lacking signal coverage, such as a car, truck, or motorcycle. The normal behavior state vector is an ordered dataset (i.e., a vector) used to mathematically and structurally describe the multi-dimensional dynamic driving characteristics of the target object at the time of acquisition. It contains key parameters that reflect the driver's immediate intention and the vehicle's physical motion trend. For example, the vector may consist of multiple components such as [instantaneous vehicle speed, acceleration, lane number, turn signal status (left / right / off), real-time steering wheel angle, and vehicle yaw rate]. These data together depict the normal driving profile of the vehicle before entering the blind spot decision point.

[0076] Specifically, this step is executed at the moment when it is predicted or actually detected that the target object in the blind spot is about to touch the boundary of the signal blind spot, because the vehicle's behavior at this moment can most directly reflect the driver's final intention for the upcoming path selection within the blind spot.

[0077] The central processing unit of the vehicle terminal continuously monitors its own GPS positioning information and pre-loaded electronic map data. When it detects that the distance between the current coordinates and the boundary of a signal blind zone is less than a preset threshold (wherein, the preset threshold is a critical spatial distance pre-configured within the vehicle terminal system to define the triggering of the pre-acquisition state, usually set within a reasonable range based on experience and experimental testing), it enters the pre-acquisition state. In this state, it polls the CAN bus at a preset extremely high frequency (wherein, the preset extremely high frequency is much higher than the data sampling rate of the CAN bus in the normal monitoring state, and is pre-calibrated according to the vehicle bus protocol, processor performance, and the physical characteristics of the required signals) to obtain the vehicle's low-level dynamic data, such as speed, acceleration, and steering wheel angle. At the same time, it reads the status of the headlight control module to determine whether the turn signals are turned on. At the precise moment when the calculated point of the target object's physical location (usually estimated by combining GPS and inertial navigation unit IMU) coincides with the blind zone boundary, the acquisition command is executed to solidify all monitored parameter values ​​at this moment and organize them into a multi-dimensional normal behavior state vector according to a predefined order and format.

[0078] S204. Based on the normal behavior state vector, the corresponding path selection confidence is calculated for each optional bifurcation path in the signal blind zone through a preset path selection probability model.

[0079] The preset path selection probability model refers to a mathematical algorithm model that has been designed, trained, and verified before system deployment, capable of establishing a probabilistic mapping relationship between vehicle behavior state and path selection behavior. For example, it can be a complex system containing a large amount of knowledge such as "if the vehicle is in the left lane and has its left turn signal on, there is a 95% probability of turning left." Optional branch paths represent, based on GIS road network data, all different driving path branches that comply with traffic rules and are available for the vehicle to choose within the specific signal blind zone that the target object is about to enter. For example, at an intersection, there are three optional branch paths: left turn, straight, and right turn. Path selection confidence refers to the quantitative evaluation value output by the model of the probability that the target object will choose each optional branch path. It is usually expressed as a probability value or percentage between 0 and 1. The sum of the confidence of all paths should theoretically be equal to 1 or 100%.

[0080] Specifically, once the normal behavior state vector is successfully collected, it is used as input to invoke the preset path selection probability model deployed in the background to perform inference calculations.

[0081] This pre-defined model is built offline by learning from massive amounts of historical vehicle trajectory data (including their state vectors before entering various intersections and their final actual path choices). The model learns the complex, non-linear relationships between states and choices. For example, the model may find that the feature combination of "vehicle speed above 60 km / h and in the middle lane" is strongly correlated with the "go straight" choice, while "vehicle speed below 30 km / h and in the rightmost lane" is strongly correlated with "turn right".

[0082] Once the real-time normal behavior state vector is input into the model, the internal algorithms (such as node identification in decision trees, weighted summation and activation function calculation in neural networks, or probabilistic inference in Bayesian networks) begin to operate. Each component of the input vector is used as a feature to locate the input within the learned high-dimensional decision space, and the model outputs the probability that the input sample belongs to each predefined path category (left turn, straight, etc.). For example, if the input vector is [speed: 25, lane: right, turn signal: right flashing], the model might output: {left turn confidence: 0.05, straight turn confidence: 0.10, right turn confidence: 0.85}. This output represents the path selection confidence for each possible branch path.

[0083] S205. Based on route selection confidence and macro traffic data, perform parallel simulations on map components.

[0084] Parallel inference refers to a computational paradigm that simultaneously launches multiple independent simulation processes, each responsible for simulating a possible bifurcation path selection, so that all possibilities are synchronously and parallelly extrapolated forward.

[0085] Specifically, this step is implemented as the initiation command for a multi-tasking, data-driven, dynamic simulation process. First, based on the confidence list output by S204, all bifurcation paths with confidence levels greater than a preset confidence threshold (a lower limit pre-configured in the system to optimize computational resources and automatically filter out bifurcation paths with extremely low probability and no practical simulation value) are identified. For each selected path, an independent simulation instance or simulation thread is created. Next, before starting these simulation instances, a query request is initiated to an external macro-level traffic data service platform via an API interface. The query content is the real-time traffic conditions related to these bifurcation paths to be simulated. For example, querying the current average traffic speed of the left-turn lane and the straight-ahead main road within the blind spot.

[0086] Then, each simulation instance is bound to its corresponding path selection confidence level and the relevant macro-traffic data retrieved. The confidence level determines its weight or priority in visualization, while the macro-traffic data directly affects the speed model for its internal trajectory calculation. Finally, a command to start parallel simulation is issued, and all created simulation instances simultaneously begin running in the background logic of the map component, entering their own independent and continuous calculation and rendering loops.

[0087] S206. Select the main path with the highest confidence value along the path, continuously calculate and render the main inference position and determine the corresponding main estimated departure time.

[0088] Among these, the highest confidence value for path selection refers to the path with the largest confidence value among all paths output from S204, representing the most likely scenario predicted by the model; the main path refers to the specific optional branching path corresponding to this highest confidence value; continuous calculation indicates that this process is not completed all at once, but is a dynamic simulation process that is repeatedly iterated at a fixed time step. Rendering refers to the action of converting the calculated geographic coordinates into a visible graphic element (such as an icon) on the map component and drawing or updating it; the master inferred location is used to represent the calculated geographic coordinates of the target object on the main path at any time during the simulation process; the master predicted departure time refers to the predicted time point at which the virtual target object completely leaves the boundary of the signal blind zone, calculated through the simulation of the entire main path.

[0089] Specifically, the simulator places a virtual object at the starting point of the main path (i.e., the blind spot entrance), with an initial speed equal to the actual entrance speed collected by S203. The simulation time t is initialized to 0. The simulator enters a loop, with each loop representing a tiny time increment Δt (e.g., 0.1 seconds). At the beginning of each loop, the simulator first obtains the GIS attributes (such as the legal speed limit) of the road segment where the virtual object is currently located. Then, it queries the real-time "macro traffic data" matching that road segment to obtain the average speed of the current road segment. The effective speed V_eff of the virtual object is set to the smaller of these two speeds, i.e., V_eff = min(legal speed limit, macro average speed). This ensures that the simulation neither exceeds the speed limit nor fails to reflect the actual impact of congestion. According to the kinematic formula ΔS = V_eff * Δt, the distance traveled within this time step is calculated. The simulator advances the position of the virtual object forward by a distance ΔS along the geometry of the main path (usually a broken line or curve), obtaining new coordinates, which are the main simulation position at the current moment. The new main simulation position coordinates are sent to the front-end map component for rendering. Simultaneously, the simulation time t is updated to t + Δt. Check if the new position has passed the end of the main path (i.e., the blind spot exit). If not, return to step 3 and continue the loop. If it has passed, the loop terminates. The actual time at this moment, plus the total simulation time t, is the final determined estimated departure time of the main path.

[0090] S207. Along all paths other than the main path, select the secondary path corresponding to the confidence level, continuously calculate and render the secondary simulation position, and determine the secondary estimated departure time.

[0091] Among them, the secondary path is the path with the selected confidence level of all paths other than the primary path, representing alternative solutions with low probability but objective existence; the secondary simulation position and the secondary estimated departure time are conceptually exactly the same as the primary simulation position and the primary estimated departure time, and are the results obtained by independent simulation calculation specifically for a certain secondary path.

[0092] Specifically, if there are N sub-paths, then N independent simulation instances will be launched simultaneously to execute this step. The theoretical implementation of this step is completely consistent with S206, both being time-based kinematic simulations constrained by macroscopic traffic data. The key difference and value lies in the fact that for each sub-path, a complete simulation process isolated from the main path and other sub-paths will be executed.

[0093] Each sub-path simulation independently queries and uses macroscopic traffic data for its own road segment. Each sub-path simulation instance maintains its own independent simulation clock, current position, current speed, and other state variables. Ultimately, each sub-path outputs a unique sub-simulated position sequence and an independent sub-estimated departure time.

[0094] In some embodiments, when there are available internal data nodes within the blind zone that can provide feedback on the target's passage, the model can be dynamically modified by using real data from intermediate processes based on data fusion and path pruning mechanisms of internal verification points, thereby improving the accuracy of long-term predictions.

[0095] Specifically, firstly, internal verification points with programmable data interfaces that may exist within the blind spot are pre-marked in the electronic map data. These verification points are physical or virtual facilities that can confirm the passage of a specific vehicle at a specific time, such as ETC gantries on highways, surveillance cameras at specific intersections with license plate recognition capabilities, or access control gate systems in specific parks. After steps S205 and S206 generate the geometric trajectories of the main path and each secondary path, each path is traversed to check whether it crosses any of the marked internal verification points. If a path (whether a main path or a secondary path) contains a verification point, the path is specially marked as a verification path, and the estimated arrival time at the verification point (ETA-VP) is calculated based on the current dynamic model of the path.

[0096] A dedicated monitor for the data interface of this verification point is activated within a preset monitoring time window. This time window is typically centered on ETA-VP, extending forward and backward by a preset time margin (referring to the flexible buffer time reserved on both sides of the expected arrival time to compensate for minor deviations between the projected speed and the actual speed, as well as data transmission delays). Its core purpose is to tolerate minor speed errors in the prediction model and potential data transmission delays, ensuring that verification opportunities are not missed due to slight time deviations. Within this time window, the monitor continuously listens to or actively queries the data interface of the verification point, attempting to detect whether there is passability verification data matching the identification of the target object in the blind spot (such as a license plate number).

[0097] If positive verification data is successfully received within the time window, path pruning and model calibration are performed. First, the verification path is determined to be the true trajectory of the target object, and all other parallel inference paths (including previous main paths or other secondary paths) and their associated computation threads and visualization elements are immediately terminated and removed from memory. This pruning action frees up significant CPU and memory resources, reducing the complexity of subsequent calculations. Second, using the geographical coordinates of this internal verification point and the timestamp of the received verification data as new, error-free initial conditions, the inference model of the verification path is calibrated, completely eliminating all prediction errors accumulated from entering the blind zone to this verification point. Subsequently, starting from this calibration point, the inference of subsequent road segments continues.

[0098] If no verification data is received by the time the monitoring window closes, it indicates that the target object most likely did not choose this path. Therefore, the confidence score for this verification path is penalized and reduced to a preset, near-zero, extremely low value (a pre-configured lower limit for confidence scores indicating an "extremely low probability" state), effectively marginalizing it in subsequent resource allocation and visualization. To maintain the conservation of the total probability, the reduced confidence score for this path is distributed to all other paths still in parallel simulation according to a specific algorithm (such as a weighted average based on existing confidence scores or a direct average), thus making the confidence distribution of the entire prediction system closer to reality.

[0099] The above technical steps achieve crucial convergence by actively utilizing sparse but high-value deterministic information within the blind zone, amidst the trend of uncertainty accumulating over time: either by confirming a path to eliminate all other possibilities and directly prune redundant calculations, thus improving computational efficiency; or by rejecting a path to dynamically adjust the probability distribution, ensuring that the prediction focus remains on the more probable scenario, thereby improving the accuracy of the prediction results.

[0100] S208. Display all primary simulation positions, primary estimated departure times, secondary simulation positions, and secondary estimated departure times in parallel.

[0101] Among them, all main simulation positions, main estimated departure times, secondary simulation positions, and secondary estimated departure times refer to a dynamic data set that encompasses all calculation results generated by all parallel simulation tasks in the preceding steps (S206 and S207) at any given time. Parallel visualization presentation refers to an information display design paradigm that synchronously, non-obstructingly, and visually distinctly displays multiple related but independent information flows (referring to the simulation results of the main path and all secondary paths) on the same interface (map component) and the same time dimension.

[0102] Specifically, once the parallel simulation begins in the backend, the user interface is refreshed dynamically in real time as the backend computation data is continuously updated. This transforms the complex results of the backend parallel computation into situational intelligence that monitoring personnel can instantly perceive and understand. Identity mapping maps the data attribute of path category (primary path / secondary path) to graphical styles. For example, the primary simulation location is rendered as a solid, brightly colored icon, while all secondary simulation locations are rendered as semi-transparent, smaller icons or icons with dashed outlines. Similarly, the primary path trajectory can be drawn with solid lines, and the secondary path trajectory with dashed lines.

[0103] Probability mapping can map the numerical attribute of path selection confidence to a visual saliency variable. For example, the higher the confidence level of a secondary path, the lower the transparency of its icon (the closer it is to the actual object); conversely, the lower the confidence level of a path, the more transparent its icon can be, serving only as a subtle cue.

[0104] Spatiotemporal mapping maps the two core spatiotemporal data points—the projected location (geographic coordinates) and the estimated departure time (timestamp)—to two-dimensional coordinates on the map canvas and text labels attached to icons, respectively. The entire process is a dynamic loop: the front-end user interface continuously listens for the latest projected data stream pushed or actively retrieved from the back-end. Upon receiving a new data packet, the front-end rendering engine immediately parses the data and, according to the aforementioned visual encoding rules, updates the attributes (location, style, content) of each corresponding graphic element (icon, path line, text label) on the map. This creates a dynamic visual effect of multiple virtual targets moving in parallel along different paths at different speeds within the blind spot, providing monitoring personnel with a comprehensive, quantitative, and intuitive insight into future uncertainties.

[0105] S209. Use the inferred location to update the display of the icon of the blind spot target object on the map component, and at the same time change the visual style of the blind spot icon corresponding to the blind spot target object to the preset blind spot rendering mode.

[0106] S210. If real-time data recovery of the target object in the blind spot is not obtained after the expected departure time, a timeout alarm message is generated and displayed on the large screen.

[0107] S211. When real-time data recovery of the target object in the blind zone is detected, switch the rendering method of the blind zone icon back to normal mode and calibrate it to the latest real-time position.

[0108] Steps S209-S211 and Figure 1 Steps S104-S106 in the illustrated embodiment are similar and can be found in the descriptions of steps S104-S106, which will not be repeated here.

[0109] In some embodiments, when the data of the target object in the blind spot is recovered, if there is a significant deviation between the actual location and the final location inferred by the system, a trajectory correction mechanism based on path backtracking and replanning is triggered to ensure the continuity and logical rationality of the historical trajectory, thereby improving the realism of the visualization effect.

[0110] Specifically, when the real-time data stream (such as GPS signal) of the target object in the blind zone is recovered, the geographical deviation is calculated first. This deviation is defined as the straight-line spatial distance between the first true geographical location of the target object after signal recovery and the end position of the main path or the final convergence path with the highest confidence in the current parallel simulation. This deviation value is then compared with a preset deviation threshold.

[0111] Once the geographic deviation is determined to be greater than a threshold, the complete historical trajectory of the target object (from the entry point into the blind zone to the current trajectory point) rendered on the map component is traversed. By calculating the geographic distance between each coordinate point on the historical trajectory and the actual restored location, the closest historical trajectory point is located and defined as the trajectory deviation point. The theoretical basis for this step is that this closest point probabilistically best represents the critical point where the prediction begins to err; that is, before this point, the trajectory basically matches the actual trajectory, while after this point, due to a driving decision not captured by the model (such as an unexpected turn), the two begin to diverge.

[0112] After determining the trajectory deviation point (starting point) and the actual recovery location (end point), the underlying path planning engine of the map component is invoked. Utilizing its detailed A*, Dijkstra, or other efficient pathfinding algorithms, and based on real underlying road network data with topological and traffic rule constraints, one or more of the most reasonable and passable paths between the starting point and the end point are planned, and the optimal one is selected as the correction path segment.

[0113] Finally, a grafting operation is performed, deleting all portions of the original historical trajectory from the deviation point to its end. The newly generated corrected path segment is then seamlessly connected after the deviation point, forming a completely new, logically consistent, and complete historical trajectory. After repairing the historical trajectory, the position of the blind spot icon representing the target object is instantly calibrated from the original projection end to the accurate real-time position after signal recovery.

[0114] The aforementioned technical steps provide a reasonable post-event explanation by tracing back to the most likely deviation point and reconstructing the path based on real road network data. This ensures that the entire historical trajectory from entering the blind spot to signal recovery remains continuous and compliant with traffic rules. This not only improves the smoothness and realism of the visualization interface but also enhances users' long-term trust in the system's predictive capabilities.

[0115] In the above embodiments, a multi-path parallel extrapolation mechanism is introduced, employing a monitor verification mechanism based on verification points and ETA. This mechanism proactively acquires external truth data at critical path bifurcation points and uses this data to dynamically adjust the confidence level of each path in real time, effectively eliminating erroneous paths. Through this closed-loop feedback combining verification and falsification, continuous iterative correction and rapid pruning of the predicted path are achieved, ultimately allowing the extrapolation results to quickly converge to the most probable true path, thus improving the accuracy and reliability of blind zone trajectory prediction.

[0116] The following describes an exemplary custom map large screen generation system 300 provided in the embodiments of this application. Figure 3This is an exemplary hardware structure diagram of the custom map large screen generation system 300 provided in this application embodiment.

[0117] In some embodiments, the custom map large-screen generation system 300 is a computer device or includes a computer device in the system. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores data. The network interface of the computer device is used to communicate with other external terminals or servers via a network connection. In some embodiments, the network interface can be a wired network interface; in some embodiments, it can also be a wireless network interface. When the computer program is executed by the processor, it implements the methods in the embodiments of this application.

[0118] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0119] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0120] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0121] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0122] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A self-defined map large screen generation method, characterized in that, The method comprises the following steps: acquiring real-time data of one or more target objects bound to a map component for generating a large screen; the data of the target objects includes their real-time positions, and icons of the target objects are rendered on the map component according to the real-time positions; when the real-time position of a target object is interrupted and the last acquired real-time position is located within a preset signal blind area geofence, recording the last valid speed, interruption position and interruption time of the corresponding blind area target object before the information interruption; determining a deduced speed according to macroscopic traffic data, and continuously calculating the deduced position and estimated exit time of the blind area target object within the signal blind area in combination with a signal blind area path of the signal blind area; the macroscopic traffic data is obtained by repeatedly requesting a third-party traffic data service interface based on the geographical position information of the signal blind area at a preset time period; updating the display of the icon of the blind area target object on the map component using the deduced position, and changing the visual style of the blind area icon corresponding to the blind area target object to a preset blind area rendering mode; when the real-time data of the blind area target object is not recovered after the estimated exit time, generating and displaying an overtime alarm information on the large screen; when the real-time data of the blind area target object is detected to be recovered, switching the rendering method of the blind area icon back to a normal mode and calibrating to the latest real-time position.

2. The method of claim 1, wherein, The method of determining a deduced speed according to macroscopic traffic data, and continuously calculating the deduced position and estimated exit time of the blind area target object within the signal blind area in combination with a signal blind area path of the signal blind area comprises the following steps: pre-dividing the signal blind area path into a preset number of ordered path sub-segments in map data; real-time maintaining a historical passing queue composed of a preset number of similar target objects that have successfully exited the signal blind area, and recording the actual total passing time of each similar target object in the queue; based on the actual total passing time, calculating a reference deduced time consumption by a weighted average algorithm, and determining a final deduced total time consumption in combination with the macroscopic traffic data; allocating the final deduced total time consumption to each path sub-segment according to a preset speed distribution model to generate an independent deduced sub-speed for each path sub-segment; the speed distribution model is a historical statistical model describing the speed change trend of target objects on different path sub-segments; when the blind area target object enters the signal blind area, the deduced position is calculated in a segmented manner according to the corresponding deduced sub-speed in sequence until the travel trajectory covers all path sub-segments, and the estimated exit time is determined according to the final deduced total time consumption.

3. The method of claim 2, wherein, The method of calculating a reference deduced time consumption based on the actual total passing time by a weighted average algorithm, and determining a final deduced total time consumption in combination with the macroscopic traffic data comprises the following steps: at the entrance of the signal blind area, real-time identifying a traffic cluster formed by vehicles entering continuously by monitoring the time interval of the vehicles entering, and defining a plurality of target objects in the same traffic cluster as a dynamic queue of the traffic cluster with internal order; determining an order position in a dynamic queue of a corresponding traffic cluster when the blind area target object enters the signal blind area; when performing the weighted average algorithm, introducing an order position weight factor to give a first calculation weight to an actual total travel time of a historical target object in the historical travel queue and having the same order position as the blind area target object in the corresponding traffic cluster; the first calculation weight is the maximum value of all weights in the weighted average algorithm; calculating a final deduced total travel time based on the weighted average algorithm with the order position weight factor.

4. The method of claim 1, wherein, The deduced speed is determined according to the macro traffic data, and the deduced position and the estimated exit time of the blind area target object in the signal blind area are continuously calculated based on the signal blind area path of the signal blind area, which specifically includes: collecting a normal behavior state vector before the blind area target object enters the signal blind area; based on the normal behavior state vector, calculating a corresponding path selection confidence for each selectable branch path in the signal blind area through a preset path selection probability model; based on the path selection confidence and the macro traffic data, performing parallel deduction on the map component: along the main path with the highest path selection confidence value, continuously calculating and rendering the main deduction position and determining the corresponding main estimated exit time; along all the secondary paths corresponding to the path selection confidence except the main path, continuously calculating and rendering the secondary deduction position and determining the secondary estimated exit time; all the main deduction positions, main estimated exit times, secondary deduction positions and secondary estimated exit times are presented in parallel.

5. The method of claim 4, wherein, After continuously calculating and rendering the secondary deduction position and determining the secondary estimated exit time along all the secondary paths corresponding to the path selection confidence except the main path, it further includes: when the secondary path or the main path contains a preset internal verification point, calculating the estimated arrival time of the blind area target object at the verification point on the verification path containing the internal verification point; monitoring the data interface of the internal verification point to detect whether there is passing verification data of the blind area target object within a preset monitoring time window before and after the estimated arrival time at the verification point; if yes, terminate and clear the parallel deduction and visualization of all other paths except the verification path, and calibrate and continue the deduction of the verification path based on the internal verification point; if no, dynamically reduce the path selection confidence of the verification path to a preset extremely low value, and increase the confidence of all other paths except the verification path by an average allocation ratio.

6. The method of claim 1, wherein, Before switching the rendering method of the blind area icon back to the normal mode and calibrating to the latest real-time position when the real-time data of the blind area target object is detected to be recovered, it further includes: In a case where the geographical deviation is greater than a preset deviation threshold, a trajectory deviation point closest to the actual recovery position in geographical distance is located in a historical inference trajectory belonging to the blind area target object that has been rendered on the map component; the geographical deviation is a position difference between the actual recovery position and the current inference position of the blind area target object; Based on the underlying road network data of the map component, a connection path between the trajectory deviation point and the actual recovery position is planned, and a correction path segment is generated; The part of the historical inference trajectory from the trajectory deviation point to the end is replaced by the correction path segment, and the blind area icon is calibrated to the latest real-time position.

7. The method of claim 1, wherein, In a case where the real-time data recovery of the blind area target object is still not acquired when the expected driving-out time is exceeded, after the timeout alarm information is generated and displayed on the large screen, the method further includes: When a neighboring signal blind area is searched within a preset geographical radius with the preset driving-out point of the signal blind area as the center, the visual style of the timeout alarm information of the blind area target object in the signal blind area is changed to a non-active mode representing that the inference is completed but the state is unknown; At the entrance of the neighboring signal blind area, a high-uncertainty speculation icon is generated, and new blind area alarm information is attached.

8. A self-defined map large screen generation system, characterized in that, The self-defined map large screen generation system includes one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code including computer instructions; the one or more processors invoke the computer instructions to enable the self-defined map large screen generation system to perform the method in any one of claims 1-7.

9. A computer program product comprising instructions, characterized in that, When the computer program product runs on the self-defined map large screen generation system, the self-defined map large screen generation system is enabled to perform the method in any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, wherein: When the instructions run on the self-defined map large screen generation system, the self-defined map large screen generation system is enabled to perform the method in any one of claims 1-7.

Images