A warehouse intelligent WMS platform freight vehicle positioning and trajectory analysis method

Abstract

The application discloses a warehouse intelligent WMS platform freight vehicle positioning and track analysis method, belongs to the warehouse logistics informatization technical field, and the method comprises the following steps: a cloud end carries out time stamp unified conversion and coordinate system unified conversion on original position messages reported by multi-brand heterogeneous vehicles, generates a standardized space-time semantic object, a front-end rendering engine carries out collision detection on candidate motion tracks of vehicle graphic objects in combination with warehouse logistics topological constraints, solves legal adsorption point correction target coordinates and carries out smooth interpolation when detecting that the vehicle graphic objects pass through obstacles, and differentiates redrawing frequency control and multi-level aggregation rendering on vehicle graphic objects inside and outside a viewport according to a quadtree space index. The application eliminates heterogeneous vehicle data islands, eliminates visual distortion phenomena such as vehicle track passing through obstacles and position mutation, and guarantees low delay and frame rate stability of front-end rendering in a large-scale concurrent scene, and can be used for warehouse vehicle positioning scheduling and track visualization analysis.

Description

Technical Field

[0001] This invention relates to the field of warehousing and logistics information technology, specifically to a method for positioning and trajectory analysis of freight vehicles applied to a warehouse intelligent WMS platform. In particular, it relates to a technical solution that can eliminate data silos of positioning of multi-brand heterogeneous freight vehicles, realize semantic-level graphic correction of positioning drift in the front-end graphical interface, and support low-latency visualization rendering of large-scale concurrent vehicle trajectories. It can be applied to the field of Internet cloud computing warehouse vehicle positioning and scheduling, as well as vehicle monitoring and scheduling analysis scenarios in the intelligent warehousing and industrial Internet industries. Background Technology

[0002] With the rapid development and popularization of Internet cloud computing technology, large-scale intelligent warehouse management systems are evolving towards a high degree of cloudification and automation. In modern warehousing and logistics operations, it has become common for freight vehicles and automated guided vehicles from different manufacturers and brands to operate together. The positioning technologies used by various vehicles cover a variety of methods, including ultra-wideband positioning, radio frequency identification positioning, visual real-time positioning and map building, and dead reckoning positioning based on inertial sensors. Their positioning accuracy, data update frequency, and communication protocol formats vary greatly, which brings significant challenges to the unified scheduling and visualization of warehouse management systems.

[0003] Existing warehouse management system platforms typically employ the following approach for vehicle positioning and trajectory analysis: a cloud server receives raw positioning coordinate data reported by various vehicles; for positioning methods such as dead reckoning that accumulate errors, numerical algorithms such as Kalman filtering are used to filter and correct the coordinate sequence in the cloud; the processed coordinate data is then sent to the front end via an interface; the front-end rendering engine draws the positions of vehicle graphic objects sequentially according to the received coordinates; and the vehicle icon coordinates are updated synchronously with the user's zooming and panning operations on the monitoring map. Existing technologies generally believe that the accuracy of vehicle graphic rendering should be synchronized with the actual coordinate updates frame by frame / technical bias, failing to recognize the ineffective consumption of rendering performance by the high-frequency coordinate updates of vehicles outside the viewport. This application breaks through this conventional understanding and proposes a differentiated strategy that maintains high-frequency rendering only for objects within the viewport and degrades objects outside the viewport to state synchronization, significantly reducing the overall rendering load without sacrificing the display accuracy of the user's visible area.

[0004] The existing technical solutions have the following shortcomings: First, due to the different positioning technologies and communication protocols used by various heterogeneous vehicles, the reported location messages lack unified standards in terms of timestamp references, coordinate system definitions, and field formats. When the cloud system performs unified visualization and scheduling of vehicles from multiple brands running in the same field, it is difficult to directly process the original messages in a unified manner, which can easily lead to data silos. Second, the existing error correction methods only stay at the cloud numerical filtering level and do not link with the front-end graphic display logic and warehouse logistics topology constraints. When numerical filtering cannot completely eliminate accumulated errors, the front-end rendering engine can only draw directly based on coordinates, which can easily lead to visual distortions on the monitoring interface, such as vehicle graphic objects crossing shelves or walls, or sudden changes in position, which are not in line with physical common sense and affect the judgment of dispatchers. Third, the existing solutions adopt the same full-scale redraw strategy for vehicles inside and outside the viewport, without differentiating the rendering load according to the user interaction range and map zoom level. When the number of vehicles reaches a large scale and the coordinates are reported at high frequency, the calculation and drawing load of the front-end rendering engine increases significantly, which can easily lead to rendering delays, frame drops, or even browser malfunctions.

[0005] Therefore, how to achieve a unified spatiotemporal representation of heterogeneous vehicle location data in the cloud, how to deeply couple positioning error correction with front-end graphics display logic to eliminate visual distortion, and how to ensure low latency and stable frame rate of front-end rendering in large-scale vehicle concurrency scenarios are technical problems that urgently need to be solved in this field. Summary of the Invention

[0006] The technical problems to be solved by this invention include: First, how to eliminate the data silos caused by different positioning technologies and communication protocols of multi-brand heterogeneous freight vehicles and automated guided vehicles on the cloud warehouse management system platform, and realize the unified spatiotemporal expression of vehicle location data; Second, how to deeply couple the positioning error correction with the front-end graphical user interface visualization display logic, and eliminate visual distortion phenomena such as vehicle trajectory crossing obstacles and sudden position changes that cannot be solved by existing cloud numerical filtering methods; Third, how to ensure low latency and stable frame rate of front-end graphical user interface rendering in scenarios where large-scale vehicles report trajectory coordinates at high frequency.

[0007] To address the aforementioned technical problems, this invention provides a method for locating and analyzing the trajectory of freight vehicles in a warehouse intelligent WMS platform. This method includes three processing stages that work together in a coordinated manner.

[0008] In the first processing stage, the cloud data access layer continuously receives raw location messages reported by multi-brand freight vehicles and automated guided vehicles. These raw location messages include device identifiers, local coordinates, local timestamps, and positioning technology type identifiers. Based on the positioning technology type identifier, the layer matches the corresponding protocol parsing template, parses the raw location messages, and extracts the device identifier, local coordinate components, local timestamps, and status fields. The local timestamps are then converted to a unified clock reference to obtain standardized timestamps. The local coordinates are then converted to a unified three-dimensional coordinate system to obtain coordinate components in the unified coordinate system. The standardized timestamps, the coordinate components in the unified coordinate system, and the normalized and encapsulated status attribute sub-objects are combined and encapsulated into a standardized spatiotemporal semantic object, written into the cloud unified data bus, and sent to the front-end rendering layer. This processing stage eliminates the data silo problem caused by differences in heterogeneous vehicle positioning technologies and communication protocols, providing a consistent data foundation for subsequent scheduling and visualization.

[0009] In the second processing stage, the front-end rendering engine receives a standardized spatiotemporal semantic object from the cloud, obtains the target coordinates of the vehicle in the current frame, and reads locally cached warehouse logistics topology constraint data. This data includes geometric descriptions of obstacles such as shelf outlines, drivable passage boundaries, and walls. The engine then locates the previous frame rendering coordinates of the corresponding graphic object in the scene graph, which, together with the target coordinates, constitutes a candidate motion trajectory line segment. Collision detection is performed between this candidate trajectory line segment and the obstacle boundaries in the warehouse logistics topology constraint data. If an intersection is detected between the candidate trajectory line segment and the obstacle boundary, the collision is considered within the drivable passage where the obstacle is located. Within the channel range, the legal point inside the boundary with the smallest Euclidean distance to the target coordinates is found and used as the legal snap-in point. As the corrected target coordinates, semantic-level affine transformation correction of the graphic object is achieved. The graphic object is smoothly interpolated from the rendering coordinates of the previous frame to the rendering target coordinates, and the interpolated rendering coordinates are updated to the attributes of the vehicle graphic object in the scene graph. Through this processing stage, the positioning error correction is pushed down to the front-end rendering layer and combined with the warehouse logistics topology constraints. This can eliminate visual distortion phenomena such as vehicle graphic objects crossing obstacles and sudden position changes caused by network latency and sensor cumulative errors, and improve the physical rationality and credibility of trajectory display.

[0010] In the third processing stage, the front-end rendering engine constructs a quadtree spatial index structure based on the warehouse floor plan map, recursively dividing the map into multiple sub-region nodes. Each sub-node maintains a set of vehicle graphic objects falling within that region. It continuously acquires the current viewport boundary range corresponding to user interactions on the graphical user interface, traverses the quadtree spatial index structure, and determines the intersection between the spatial region covered by each node and the current viewport boundary range. Vehicle graphic objects in nodes intersecting with the viewport are selected as in-viewport objects, while those in other nodes are considered out-of-viewport objects. The engine calculates the aggregated rendering level based on the ratio between the current viewport boundary range and the total area of ​​the warehouse floor plan map. The rendering engine then reaches a predetermined level. With a threshold set, multiple adjacent vehicle graphic objects within the viewport are merged and drawn into a single aggregate icon, and the number of vehicles represented is marked on the icon. The coordinates of objects within the viewport are updated and redrawn at a frequency matching the device refresh rate, while objects outside the viewport are redrawn at a preset frequency lower than this frequency, maintaining only necessary state synchronization. Before each frame is rendered, the rendering time interval is controlled by combining anti-jitter and throttling mechanisms. Through this processing stage, in scenarios where large-scale vehicles concurrently report coordinates, only vehicles within the viewport are redrawn at a high frequency, while vehicles outside the viewport and distant vehicle clusters are downgraded for refresh and aggregate display, thereby ensuring the stability of the rendering frame rate and achieving low-latency dynamic scheduling analysis.

[0011] The three processing stages mentioned above work together. The first processing stage provides a unified data foundation, the second processing stage ensures the accuracy and physical rationality of the display, and the third processing stage ensures the display performance in large-scale scenarios. Together, they form a complete closed-loop technical solution.

[0012] As an optional implementation, the specific method for performing unified clock reference conversion on the local timestamp is as follows: The local timestamp is added to a pre-calibrated clock offset compensation amount obtained through network time protocol synchronization to obtain the standardized timestamp. The conversion relationship is as follows: ,in, This represents the converted, standardized timestamp. This represents the original timestamp reported locally by the device. This indicates the amount of clock offset compensation between the device and the cloud standard clock.

[0013] As an optional implementation, the specific method for performing a unified three-dimensional coordinate system transformation on the local coordinates is as follows: using rotation matrices and translation vectors pre-calibrated for each device type, the coordinate components in the local coordinate system are transformed, and the transformation relationship is as follows: ,in, , , This represents the coordinate components in the unified coordinate system after transformation. , , This represents the local coordinate system coordinate components reported by the device. This represents the rotation matrix of the device's local coordinate system relative to the unified coordinate system. This represents the corresponding translation vector.

[0014] As an optional implementation, the specific method for collision detection between the candidate trajectory segment and the obstacle boundary is as follows: calculate the cross product of the direction vector of the candidate trajectory segment and the direction vector of the obstacle boundary segment to be detected, and combine the sign of the relative positions of the endpoints of the two segments to determine whether the candidate trajectory segment crosses the obstacle boundary. The determination relationship is as follows: ,in, , Let represent the start and end vectors of the candidate motion trajectory segment, respectively. , Let represent the start and end vectors of the boundary line segment of the obstacle to be detected, respectively. This is the result of the cross product of the direction vectors.

[0015] As an optional implementation, the specific method for smooth interpolation transition of graphic objects is as follows: the interpolation coefficients are linearly increased from zero to one with the rendering frame sequence to achieve a continuous and smooth transition of the vehicle graphic object's position. The interpolation relationship is as follows: , ,in, This represents the intermediate rendering coordinates during the interpolation process. This indicates the rendering coordinates of the previous frame. Indicates the coordinates of the rendering target. These are the interpolation coefficients.

[0016] As an optional implementation, the specific method for calculating the aggregate rendering level is as follows: ,in, This represents the calculated aggregate rendering level. This represents the total area of ​​the warehouse map. This indicates the actual map area covered by the current viewport. This indicates the rounding up operation.

[0017] As an optional implementation method, the specific way to control the rendering time interval is as follows: ,in, This indicates the time interval during which the next rendering is allowed. This indicates the preset target rendering frame rate. Indicates the current moment. This indicates the last time a rendering was actually performed.

[0018] As a further optimization of the spatiotemporal semantic alignment processing stage, reference tags with known coordinates can be pre-deployed in the warehousing scenario. During operation, each vehicle positioning device periodically collects the observation data of the reference tags. The cloud dynamically solves and updates the rotation matrix and translation vector used for coordinate system transformation based on the observation data to adapt to the drift of transformation relationship caused by changes in the warehousing environment or adjustments to the equipment installation position.

[0019] This invention also provides a freight vehicle positioning and trajectory analysis system for a warehouse intelligent WMS platform corresponding to the above method, including a cloud data access layer, a cloud spatiotemporal semantic alignment engine, a data distribution channel, and a WMS front-end graphical user interface rendering layer. The front-end rendering layer includes a scene graph management module, a semantic-level graph correction module, a quadtree spatial index module, and a multi-level aggregation rendering module. Each module cooperates with the other to realize the processing stages of the above method.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] 1. By using a cloud-based spatiotemporal semantic alignment engine to perform unified timestamp conversion and unified coordinate system conversion on the original location messages of heterogeneous vehicles from multiple brands, data silos can be eliminated, providing a consistent data foundation for subsequent scheduling and visualization.

[0022] 2. By correcting positioning errors and bringing them down to the front-end rendering layer, and combining them with warehousing and logistics topology constraints for collision detection, semantic-level affine transformation, and smooth interpolation, visual distortions such as vehicle graphic objects crossing obstacles and sudden position changes can be eliminated, thereby improving the physical rationality and credibility of trajectory display.

[0023] 3. By using a dynamic viewport rendering and multi-level aggregated rendering strategy based on quadtree spatial indexing, the stability of the rendering frame rate can be guaranteed in large-scale vehicle concurrency scenarios, and low-latency dynamic scheduling analysis can be achieved. Attached Figure Description

[0024] Figure 1 This is a flowchart of the cloud-based spatiotemporal semantic alignment processing of the present invention;

[0025] Figure 2 This is a flowchart of the front-end graphic semantic level correction processing of the present invention;

[0026] Figure 3 This is a flowchart of the quadtree spatial index aggregation rendering process of the present invention. Detailed Implementation

[0027] The following is in conjunction with the appendix Figure 1-3The specific embodiments of the present invention will be further described below. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0028] The warehouse intelligent WMS platform provided in this embodiment includes, from bottom to top, a heterogeneous vehicle perception layer, a cloud data access layer, a cloud spatiotemporal semantic alignment engine, a cloud scheduling and business processing layer, a data distribution channel, and a WMS front-end graphical user interface rendering layer. The heterogeneous vehicle perception layer includes multi-brand freight vehicles and automated guided vehicles (AGVs) employing different positioning technologies such as ultra-wideband positioning, radio frequency identification (RFID) positioning, visual real-time positioning and map building, and dead reckoning. Specifically, it includes ultra-wideband positioning vehicles, RFID positioning vehicles, visual positioning vehicles, and dead reckoning positioning AAVs 114. Each vehicle reports its original location message to the cloud data access layer through its own communication protocol. The cloud data access layer can be implemented based on a message queue cluster. The cloud-based spatiotemporal semantic alignment engine receives and buffers raw messages reported during high concurrency. It performs protocol parsing, timestamp unification, and coordinate system transformation on the received raw messages to generate standardized spatiotemporal semantic objects. These objects are then written into a unified data bus for use by the cloud scheduling and business processing layers. The data distribution channel distributes the standardized spatiotemporal semantic objects and warehouse physical topology constraint data to the WMS front-end graphical user interface rendering layer. The warehouse logistics topology constraint data includes geometric information of obstacles such as shelves, aisles, and walls. The front-end rendering layer includes a scene graph management module, a semantic-level graphics correction module, a quadtree spatial index module, and a multi-level aggregation rendering module. These modules work together to ultimately present the corrected and optimized vehicle positions and trajectories on the user device screen.

[0029] The cloud data access layer continuously receives raw location messages reported by various heterogeneous vehicles and automated guided vehicles. The messages enter the message access node. The raw location messages include device identifier, local coordinates, local timestamp, positioning technology type identifier, and raw status field. For each received raw location message, the protocol parsing node matches the corresponding protocol parsing template according to the positioning technology type identifier it carries, and extracts structured data such as device identifier, local coordinate components, local timestamp, and status field. For abnormal messages with missing fields or timestamps that exceed the limit, they enter the abnormal handling branch and are discarded or compensated according to preset rules.

[0030] The unified timestamp conversion node performs a unified clock reference conversion on the extracted local timestamps to obtain standardized timestamps. The conversion relationship is as follows:

[0031] ;

[0032] in, This represents the converted, standardized timestamp. This represents the original timestamp reported locally by the device. This indicates the clock offset compensation amount between the device and the cloud standard clock, either pre-calibrated or synchronized through a network time protocol. For example, if an automated guided vehicle reports a specific timestamp locally, and the pre-calibrated offset compensation amount between it and the cloud standard clock is a small positive value, then adding the local timestamp to the offset compensation amount will yield the standardized timestamp corresponding to the message. This allows messages from different devices to be sorted and compared under a unified time base.

[0033] The coordinate system transformation node performs a unified three-dimensional coordinate system transformation on the extracted local coordinates, converting the local coordinate system data used by each device to the unified three-dimensional coordinate system of the warehouse. The transformation relationship is as follows:

[0034] ;

[0035] in, , , This represents the coordinate components in the unified coordinate system after transformation. , , This represents the local coordinate system coordinate components reported by the device. This represents the rotation matrix of the device's local coordinate system relative to the unified coordinate system. This represents the corresponding translation vector. and All data are obtained through pre-calibration and stored separately for each device type. For example, for vehicles served by ultra-wideband positioning base stations whose installation direction deviates from the unified coordinate system, their local coordinate system has a certain rotational and positional offset relationship relative to the unified coordinate system. After obtaining the corresponding rotation matrix and translation vector through pre-calibration, the local coordinates reported by the vehicle can be accurately converted to the unified coordinate system, eliminating the problem of coordinate system inconsistency caused by different installation methods.

[0036] The status attribute standardization encapsulation node normalizes and encapsulates the device's operating status fields, such as speed, orientation angle, task status, and battery level, according to a predefined unified semantic field specification, forming status attribute sub-objects. The semantic object output node combines and encapsulates the standardized timestamp, unified coordinates, and status attribute sub-objects into a standardized spatiotemporal semantic object, and writes it to the cloud unified data bus for subsequent scheduling business processing and front-end distribution.

[0037] The front-end rendering engine receives standardized spatiotemporal semantic objects of vehicles from the cloud through the cloud semantic coordinate input module, obtains the target coordinates of the vehicle in the current frame, and reads locally cached warehouse logistics topology constraint data from the warehouse topology constraint database. The topology constraint data includes geometric description information of obstacles such as shelf outlines, drivable passage boundaries, and walls. The front-end scene graph management module locates the previous frame rendering coordinates of the graphic object corresponding to the vehicle in the scene graph, which together with the target coordinates received in the current frame constitute candidate motion trajectory line segments.

[0038] The collision detection module performs collision detection between candidate motion trajectory segments and obstacle boundaries in the warehouse topology constraint data, determining whether the two intersect. The intersection determination relationship is as follows:

[0039] ;

[0040] in, , Let represent the start and end vectors of the candidate motion trajectory segment, respectively. , Let represent the start and end vectors of the boundary line segment of the obstacle to be detected, respectively. The result of the cross product of the two direction vectors, combined with the sign of the relative positions of the endpoints of the two line segments, can determine whether the candidate trajectory line segment crosses the boundary of the obstacle. For example, when a vehicle's reported coordinates in this frame fall within the area enclosed by the shelf outline due to sensor cumulative error, the candidate trajectory line segment can be determined to have crossed the shelf boundary by performing a cross product operation on the line segment formed by the rendering coordinates of the previous frame and the target coordinates of this frame with the shelf outline boundary line segment.

[0041] If the collision detection module determines that the candidate trajectory line segment crosses an obstacle, the semantic-level affine transformation module 350, within the drivable channel where the obstacle is located, solves for a legal adsorption point whose distance from the target coordinates meets the preset adsorption condition based on topological constraints. This point is then used as the corrected target coordinates to achieve semantic-level affine transformation correction of the graphic object. The corrected target coordinates replace the original target coordinates and are used as the actual target position for rendering in this frame. If the collision is not determined to be a crossing, the original target coordinates are directly used as the actual target position for rendering in this frame.

[0042] The smooth interpolation output module performs a smooth interpolation transition on the graphic object from the rendering coordinates of the previous frame to the rendering target coordinates of the current frame. The interpolation relationship is as follows:

[0043] ;

[0044] in, This represents the intermediate rendering coordinates during the interpolation process. This indicates the rendering coordinates of the previous frame. Indicates the coordinates of the rendering target in this frame. The interpolation coefficients linearly increase from zero to one with the rendering frame sequence, achieving a smooth and continuous transition of the vehicle graphic object's position, eliminating visual distortion caused by coordinate jumps, and updating the rendered coordinates obtained by smooth interpolation to the corresponding vehicle graphic object's attributes in the scene graph, completing the drawing and display of this frame. There is a two-way interactive relationship between the collision detection module and the warehouse topology constraint database, that is, the collision detection module reads the latest obstacle geometry information from the topology constraint database during each frame detection process, ensuring that the collision detection results are consistent with the actual physical layout of the warehouse.

[0045] During the initialization phase, the front-end rendering engine constructs a quadtree spatial index structure based on the overall extent of the warehouse floor plan map. Starting from the root node, the map is recursively divided into multiple sub-region nodes. Each sub-node maintains a set of vehicle graphic objects that fall into that region. The current user viewport range is marked by the viewport range box in the figure. Vehicle graphic objects located within the viewport range box and vehicle graphic objects located outside the viewport range box correspond to different rendering processing paths.

[0046] The viewport extent acquisition node listens in real time to user interaction operations on the graphical user interface, such as dragging, zooming in and out of a local area, etc., to obtain the boundary extent data of the current viewport. The quadtree node traverses the quadtree index structure. The viewport inside and outside node determination node determines the intersection of the spatial area covered by each node with the current viewport boundary extent, filters out the set of nodes that intersect with the viewport, and retains only the vehicle graphic objects in this set as objects to be rendered. The vehicle graphic objects in the other nodes are marked as objects outside the viewport.

[0047] The scaling level determination node calculates the aggregate rendering level based on the ratio between the current viewport range and the total map range. The calculation relationship is as follows:

[0048] ;

[0049] in, This represents the calculated aggregate rendering level. This represents the total area of ​​the warehouse map. This indicates the actual map area covered by the current viewport. This indicates an up-rounding operation. The calculated aggregate rendering level is compared with a preset threshold. If the threshold is reached or exceeded, the aggregate rendering branch is entered, where multiple adjacent vehicle graphic objects in the viewport are merged and drawn into a single aggregate icon, and the number of vehicles represented is marked on the icon. If the threshold is not reached, the high-frequency redraw branch is entered, maintaining the conventional rendering mode of drawing each vehicle graphic object one by one. For example, when the user zooms out the monitoring map to a global view that only shows the entire warehouse area, the actual map area covered by the viewport is close to the total area of ​​the warehouse map. At this time, the calculated aggregate rendering level value is high. Once the preset threshold is reached, the densely distributed vehicle graphic objects in this area are aggregated and displayed to avoid the problem of excessive rendering load caused by drawing a large number of icons at the same time. When the user zooms in on the map to observe a local area in detail, the actual map area covered by the viewport is significantly smaller than the total map area, and the aggregate rendering level value decreases. At this time, the conventional rendering mode of drawing each vehicle graphic object in this local area is used to ensure the precision of the vehicle position display.

[0050] For vehicle graphic objects determined to be within the viewport, coordinate updates and redraws are performed at a high frequency matching the device refresh rate. For vehicle graphic objects determined to be outside the viewport, their redraw frequency is downgraded to a preset low-frequency queue, maintaining only necessary state synchronization. Before each frame is rendered, the rendering frame rate control node controls the rendering time interval using anti-jitter and throttling mechanisms. The control relationship is as follows:

[0051] ;

[0052] in, This indicates the time interval during which the next rendering is allowed. This indicates the preset target rendering frame rate. Indicates the current moment. This indicates the last time a rendering was actually performed. This control relationship helps avoid repeated invalid rendering caused by too frequent interactive operations, thus ensuring the stability of the rendering frame rate.

[0053] As an alternative implementation of the present invention, in the spatiotemporal semantic alignment stage, in addition to using pre-calibrated rotation matrices and translation vectors for coordinate system transformation, a real-time calibration method based on known reference points can also be adopted. That is, several reference tags with known precise coordinates are pre-deployed in the warehousing scenario. During operation, each vehicle positioning device periodically collects the observation data of the reference tags. The cloud dynamically solves and updates the coordinate transformation parameters based on the observation data to adapt to the drift of the transformation relationship caused by changes in the warehousing environment or adjustments to the equipment installation position, thereby achieving adaptive calibration of the coordinate system transformation relationship.

[0054] As another alternative implementation of the present invention, in the semantic-level graphics correction stage, in addition to using collision detection and legal adsorption point solving methods based on line segment intersection determination, a passable grid map pre-built based on the navigable area of ​​the warehouse can also be used. When the front-end rendering engine detects that the candidate trajectory deviates from the navigable grid, it solves a legal path connecting the previous frame position and the target position in real time within the navigable grid range through a path search algorithm, and performs segmented rendering and interpolation transition along the path. This can also achieve the invention's objective of eliminating visual distortion phenomena such as vehicle graphic objects crossing obstacles and sudden position changes, and can be used as a supplementary implementation of the semantic-level graphics correction method.

[0055] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention 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. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for locating and analyzing the trajectory of freight vehicles in a warehouse intelligent management system (WMS) platform, characterized in that, Includes the following steps: The cloud data access layer receives raw location messages reported by multi-brand freight vehicles and automated guided vehicles. The raw location messages include device identifiers, local coordinates, local timestamps, and positioning technology type identifiers. The raw location messages are parsed according to the positioning technology type identifiers to obtain device identifiers, local coordinate components, local timestamps, and status fields. The local timestamps are converted to a unified clock reference to obtain standardized timestamps. The local coordinates are converted to a unified three-dimensional coordinate system to obtain coordinate components in the unified coordinate system. The standardized timestamps, the coordinate components in the unified coordinate system, and the normalized encapsulated status attribute sub-objects are combined and encapsulated into a standardized spatiotemporal semantic object and sent to the WMS front end. The WMS front-end rendering engine receives the standardized spatiotemporal semantic object, obtains the target coordinates of the vehicle in the current frame, and reads the locally cached warehouse logistics topology constraint data. The warehouse logistics topology constraint data includes the outline of the shelves, the boundary of the drivable passage, and the geometric description information of the walls. It locates the previous frame rendering coordinates of the corresponding graphic object of the vehicle in the scene graph, and forms a candidate motion trajectory line segment with the target coordinates. It performs collision detection between the candidate motion trajectory line segment and the obstacle boundary in the warehouse logistics topology constraint data. If the candidate motion trajectory line segment intersects with the obstacle boundary, within the drivable passage where the obstacle is located, it solves for the legal point inside the boundary with the smallest Euclidean distance to the target coordinates, which is used as the legal snap point and the corrected target coordinates. It performs smooth interpolation transition on the graphic object from the previous frame rendering coordinates to the rendering target coordinates. Wherein, if the candidate motion trajectory line segment intersects with the obstacle boundary, the rendering target coordinates are the corrected target coordinates; if no intersection is detected, the rendering target coordinates are the target coordinates. The interpolated rendering coordinates are updated to the attributes of the vehicle graphic object in the scene graph. The WMS front-end rendering engine constructs a quadtree spatial index structure based on the warehouse floor plan map, obtains the boundary range of the current user viewport, and determines the intersection of the spatial area covered by each node in the quadtree spatial index structure with the boundary range of the viewport. Vehicle graphic objects in the nodes that intersect with the viewport are selected as objects within the viewport, and vehicle graphic objects in the remaining nodes are selected as objects outside the viewport. The aggregation rendering level is calculated based on the ratio between the boundary range of the viewport and the total range of the warehouse floor plan map. When the aggregation rendering level reaches a preset threshold, multiple adjacent vehicle graphic objects in the space within the viewport are merged and drawn into a single aggregation icon. The coordinates of the objects within the viewport are updated and redrawn at a frequency matching the device refresh rate, and the objects outside the viewport are redrawn at a preset frequency lower than the frequency mentioned above.

2. The method according to claim 1, characterized in that, The steps for performing unified clock reference conversion on the local timestamp include: adding the local timestamp to a pre-calibrated clock offset compensation amount to obtain the standardized timestamp, wherein the clock offset compensation amount represents the difference between the local clock of the corresponding device and the standard clock in the cloud.

3. The method according to claim 1, characterized in that, The steps for performing a unified three-dimensional coordinate system transformation on the local coordinates include: using a rotation matrix and translation vector pre-calibrated for the corresponding device type to perform a coordinate system transformation on the local coordinates to obtain the coordinate components under the unified coordinate system.

4. The method according to claim 1, characterized in that, The step of performing collision detection between the candidate motion trajectory segment and the obstacle boundary includes: calculating the cross product of the direction vector of the candidate motion trajectory segment and the direction vector of the obstacle boundary segment, and determining whether the candidate motion trajectory segment crosses the obstacle boundary by combining the sign of the relative positions of the endpoints of the two segments.

5. The method according to claim 1, characterized in that, The steps for smoothing the transition of the graphic object by interpolation include: performing linear interpolation between the rendering coordinates of the previous frame and the rendering target coordinates by linearly increasing the interpolation coefficient from zero to one, so as to obtain the intermediate rendering coordinate sequence in the interpolation process.

6. The method according to claim 1, characterized in that, The steps for calculating the aggregate rendering level based on the ratio between the boundary range of the viewport and the total range of the warehouse map include: performing a base-2 logarithmic operation with the ratio of the total area of ​​the warehouse map to the actual map area covered by the viewport as the argument, and rounding the result up to obtain the aggregate rendering level.

7. The method according to claim 1, characterized in that, The step of redrawing the object outside the viewport at a preset frequency lower than the frequency includes: adding the object outside the viewport to a preset low-frequency redraw queue, and only synchronizing its execution status without updating its coordinates frame by frame.

8. The method according to any one of claims 1 to 7, characterized in that, Before each frame is rendered, the difference between the time interval corresponding to the preset target rendering frame rate and the current time and the last actual rendering time is obtained. The larger of the two values ​​is used as the time interval for the next rendering to be executed, so as to control the rendering time interval.

9. The method according to claim 3, characterized in that, Also includes: In a warehousing scenario, reference tags with known coordinates are pre-deployed. During operation, the vehicle positioning equipment periodically collects observation data from the reference tags. The cloud dynamically solves and updates the rotation matrix and translation vector used for coordinate system transformation based on the observation data.

10. A freight vehicle positioning and trajectory analysis system for a warehouse intelligent WMS platform, characterized in that, include: A cloud-based data access layer receives raw location messages from multi-brand freight vehicles and automated guided vehicles. A cloud-based spatiotemporal semantic alignment engine, connected to the cloud-based data access layer, performs protocol parsing, timestamp unification conversion, and coordinate system unification conversion on the raw location messages to generate standardized spatiotemporal semantic objects. A data delivery channel distributes the standardized spatiotemporal semantic objects and warehousing and logistics topology constraint data to the WMS front-end GUI rendering layer. The WMS front-end GUI rendering layer includes a scene graph management module, a semantic-level graphics correction module, a quadtree spatial index module, and multiple... The hierarchical aggregation rendering module, the semantic-level graphics correction module is used to perform collision detection on the candidate motion trajectory segments of the vehicle graphic object based on the warehousing and logistics topology constraint data, and solve for the legal adsorption points that meet the preset adsorption conditions as the corrected target coordinates when intersections are detected, and perform smooth interpolation transition on the vehicle graphic object. The quadtree spatial index module and the multi-level aggregation rendering module are used to perform viewport-inside-outside division and aggregation rendering hierarchy calculation on the vehicle graphic object based on the boundary range of the user viewport, and use different redraw frequencies for objects inside and outside the viewport.

Images