A hazardous waste landfill tracking method and system based on space-time anchor points and voxelization model

Abstract

The application discloses a hazardous waste landfill tracing method and system based on a space-time anchor point and a voxel model, and relates to the field of hazardous waste treatment technology. The method comprises the following steps: establishing a three-dimensional reference model of a landfill area; capturing a "space-time anchor point" when hazardous waste is dumped; generating a three-dimensional model after the operation and calculating an incremental landfill body; dividing the incremental landfill body into structured voxels with unique indexes; searching for associated voxels through the space-time anchor point, injecting business information; and realizing visual and accurate query. The system comprises a vehicle-mounted sensing unit, a space surveying and mapping unit, a central processing and data management unit (data processing and association), and a visual interaction unit. The application breaks the traditional "batch-area" fuzzy tracing mode, realizes "batch-point" accurate tracing, improves management accuracy and efficiency, reduces the risk of environmental pollution, provides technical support for responsibility identification, and can be widely applied to the fields of hazardous waste landfill and bulk material storage management.

Description

Technical Field

[0001] This invention relates to the field of hazardous waste treatment and environmental management technology, and in particular to a method and system for tracing hazardous waste landfills based on spatiotemporal anchor points and volumetric models. Background Technology

[0002] With the increasing global awareness of environmental protection, the treatment and management of hazardous waste has become an important issue in the environmental protection field. Landfill disposal is one of the most widely used methods for hazardous waste disposal. However, in practice, how to efficiently and accurately manage and trace the landfill process has always been a challenge for the industry. Especially when pollution leaks or safety accidents occur at landfills, the lack of effective tracing methods often makes it impossible to quickly locate the source of pollution, thus affecting the subsequent determination of responsibility and pollution control.

[0003] (a) Deficiencies of existing technology

[0004] Traditional hazardous waste landfill management relies on manual recording, ledger management, and basic Geographic Information System (GIS) technology. While these methods can record the transportation, dumping, and landfilling of hazardous waste to some extent, their management system has many shortcomings:

[0005] Information fragmentation creates a "black box": Traditional management methods break down the transportation, dumping, and landfilling of hazardous waste into multiple stages, with a lack of effective correlation between the operational ledgers and spatial mapping information of each stage. Even if a pollution incident occurs within the landfill, managers find it difficult to accurately trace the specific source of pollution—for example, although they know the landfill area, they cannot determine the specific location and batch of waste dumped within that area, severely hindering the source tracing process.

[0006] Currently, spatial data for landfills is typically obtained through point cloud data or triangulation data acquired via drones, satellites, or ground-based measurements. While these unstructured geometries can provide spatial references, they cannot be directly matched with detailed business information such as waste batch numbers, source units, and waste types. This results in a lack of refined and intelligent management, making it difficult to meet the demands of modern environmental supervision.

[0007] Traditional traceability methods rely on fuzzy matching of "batch-area," which cannot achieve precise traceability of landfills. When it is necessary to locate the specific location of a batch of waste, it is only possible through inference and indirect methods, which have low accuracy and cannot provide timely data for responsibility tracing, increasing the difficulty of environmental accident investigations and reducing the efficiency of responsibility determination.

[0008] (II) Current Status of Precise Recording and Utilization of Spatiotemporal Data

[0009] With the popularization of big data, IoT, and drone technologies, landfill management is gradually developing towards digitalization and intelligence, highlighting the increasing importance of spatiotemporal data (including the three-dimensional spatial coordinates and timestamps of waste). Some existing technologies attempt to combine landfill spatial information with waste batch information, such as obtaining high-precision waste dumping location data through RTK positioning systems. However, in traditional management systems, this spatial data cannot be correlated in real time with information such as waste batches, transport vehicles, and dumping times, resulting in a lack of accuracy and timeliness in traceability. Simultaneously, unstructured spatial data (such as point clouds and triangulation networks) is difficult to combine with structured waste inventory information, leading to high data analysis difficulty, low query efficiency, and an inability to support efficient management.

[0010] (III) Needs for Intelligent Traceability and Management

[0011] The field of hazardous waste management urgently needs novel solutions to achieve precise traceability and efficient management of landfills throughout the entire process. This requires defining unique "spatiotemporal anchor points" for waste through innovative technologies, binding spatial location, time points, and batch information to break down the fuzzy "batch-region" traceability barriers; transforming unstructured spatial data into structured units (such as volumetric elements) to establish precise correlations with waste attributes; and achieving fully automated data acquisition and seamless integration with existing hardware and software. Utilizing technologies such as the Internet of Things, artificial intelligence, and digital twins to achieve real-time monitoring, precise traceability, and intelligent analysis of hazardous waste management has become an inevitable trend in the industry. Therefore, technical personnel in this field are dedicated to developing a precise traceability method for hazardous waste landfills based on spatiotemporal anchor points and structured spatial models. This method is of great significance for improving the accuracy and efficiency of landfill management, ensuring the safety and traceability of the landfill process, and providing technical support for environmental pollution prevention and control and accountability. Summary of the Invention

[0012] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is that in traditional management, the business ledger recording waste transportation and the spatial mapping recording landfill location are separated. Even if the landfill area is known, it is impossible to determine the specific batch of waste buried in the area, forming a "black box" of mixed information, making it difficult to accurately locate the source of responsibility when pollution occurs. In addition, the newly added landfill bodies obtained by UAV scanning are irregular three-dimensional geometric models (point clouds or triangular networks). This kind of unstructured data cannot be directly matched with structured business information (such as waste lists) in the database, making it difficult to manage and analyze spatial data in a refined manner.

[0013] To achieve the above objectives, this invention provides a method for tracing hazardous waste landfills based on spatiotemporal anchor points and a volumetric model:

[0014] Includes the following steps:

[0015] Step 1: System Initialization and Data Pre-binding. Before establishing the baseline 3D model, a UAV is used to perform standardized flight paths over the landfill unit to collect image data containing high-precision position and attitude information. The image data is processed using Structure from Motion (SfM) and Multi-View Stereo (MVS) algorithms to generate a 3D baseline model M1 for the landfill area. The SfM algorithm estimates the camera's position and attitude by calculating image points of the same object from different viewpoints, reconstructing the 3D point cloud and satisfying the linear relationship: p = K[R|t]P, where p is a 2D image point, P is a 3D spatial point, R and t are the camera's rotation matrix and translation vector, respectively, and K is the camera's intrinsic parameter matrix. MVS, based on SfM, refines the point cloud data by calculating pixel matching between images, increasing point cloud density and correcting geometric errors.

[0016] Step 2: Capture the spatiotemporal anchor point of the dumping event. When the transport vehicle arrives at the landfill site and begins dumping, the system captures the "spatiotemporal anchor point" data packet. Precise three-dimensional coordinates (X, Y, Z) are obtained through the vehicle-mounted RTK-GNSS device. Combined with sensor data (such as tilt sensors) or manual confirmation (such as physical buttons) triggering the event, a data packet containing the waste batch ID, timestamp, three-dimensional coordinates, and trigger source is generated. The structure is: AnchorPoint={"batch_id": "HW-2025-A001","timestamp": 1750898826123, "coordinate": (X, Y, Z), "source_type": "RTK_AUTO"}. The captured spatiotemporal anchor point data is transmitted in real-time to the backend platform via wireless network and stored in the database.

[0017] Step 3: Acquire spatial data after the operation and calculate the incremental model. After the landfill operation is completed, the UAV flies again to collect image data and generate a second-phase 3D model M2. Comparing the two phases of 3D models M1 and M2, the newly added landfill body model, i.e., the incremental model ΔV, is calculated using the following formula, and the 3D coordinates of the incremental area are obtained through geometric difference analysis:

[0018] ΔV= =

[0019] Where M2 (Xi,Yi,Zi) and M1 (Xi,Yi,Zi) represent the three-dimensional coordinate values ​​at time points T2 and T1, respectively;

[0020] Step 4: Spatial Change Calculation and Structured Voxelization. To facilitate the storage and retrieval of spatial data, the incremental model ΔV is subdivided into discrete, standard cubic units with unique address indices, i.e., voxels Vi, according to a set side length, using the formula Vi=Voxelize(ΔV). Each voxel is assigned a unique three-dimensional coordinate index (a,b,c). Octrees and kd-dimensional trees are used to optimize the efficiency of spatial data access.

[0021] Octree: The three-dimensional space is recursively divided into eight sub-regions according to the formula Node(a,b,c)=Octree(Vi). During querying, only the relevant sub-regions are accessed, reducing the amount of computation.

[0022] kd-tree: The formula Nodi=KDTree(Vi,split_dimension) generates a binary tree structure by dividing the space by dimension, which can quickly filter matching elements;

[0023] Step 5: The targeted attribute injection system based on spatiotemporal anchor points generates a 3D spatial query volume (e.g., a sphere with a radius of 5 meters) according to the coordinates of the spatiotemporal anchor points, and retrieves all voxels related to the spatiotemporal anchor points. The business information associated with the spatiotemporal anchor point (e.g., hazardous waste batch, transport vehicle information, source unit, waste type) is injected into the attribute fields of the voxels using the formula Vi=InjectAttributes(Vi,business_info), achieving precise association between the spatiotemporal anchor point and spatial data (where Vi is the retrieved voxel, and business_info is the business information associated with the spatiotemporal anchor point).

[0024] Step Six: Data Persistence and Visualization Applications store the injected attribute voxels in a spatial database and establish a query index; front-end and back-end communication is achieved through RESTful API or GraphQL. Users can click on any position of the 3D model through the front-end visualization module, and the system retrieves relevant voxel attributes from the database and displays detailed waste information in real time, completing the "source tracing by map".

[0025] In a preferred embodiment of the present invention, the hazardous waste landfill traceability system based on spatiotemporal anchor points and volumetric models is an Internet of Things and digital twin fusion system. Its architecture is divided into a perception layer, a network layer, a platform layer, and an application layer. Its physical components include an onboard perception unit, a spatial mapping unit, a central processing and data management unit, and a visualization interaction unit. The functions and structures of each unit are as follows:

[0026] Vehicle-mounted sensing unit: Installed on the hazardous waste transport vehicle, it is a hardware generator for "spatiotemporal anchor points," and its core components include:

[0027] Positioning module: Industrial-grade high-precision RTK receiver, providing real-time, centimeter-level accuracy 3D coordinates;

[0028] Triggering module: MEMS (Micro-Electro-Mechanical Systems) tilt sensor (installed on the truck bed beam, communicates with the main controller via CAN bus or RS-485 interface, with an accuracy better than 0.1 degrees) or industrial-grade physical button (installed in the driver's operating area, providing a manual trigger signal);

[0029] Main controller (MCU): An embedded microcontroller (such as an ARM Cortex-M series microcontroller) is responsible for parsing the NMEA-0183 format data output by the positioning module, listening to the trigger module signal, freezing and packaging the coordinates and timestamp, and obtaining the preset vehicle / task ID;

[0030] Communication module: Full network compatible 4G / 5G Cat-1 / Cat-4 communication module, which establishes a connection with the central processing unit through MQTT or HTTP / S protocol and reliably uploads "spatiotemporal anchor" data packets.

[0031] Spatial mapping unit: responsible for acquiring high-precision 3D scene data; core components include:

[0032] Flight platform: Industrial-grade multi-rotor UAV with RTK / PPK differential positioning capability, achieving centimeter-level flight path accuracy and ground control point accuracy;

[0033] Mission payload: Full-frame, high-resolution (45 megapixels or higher) orthophoto or oblique photography camera, equipped with a low-distortion fixed-focus lens to ensure image data accuracy.

[0034] Central Processing and Data Management Unit: A backend software system deployed on a server (local or cloud), consisting of multiple decoupled microservice modules.

[0035] Data access and gateway service: system entry point, handling MQTT connection and data reporting of vehicle-mounted sensing units, and image data upload of spatial mapping units;

[0036] 3D Reconstruction and Spatial Analysis Services: Embedded photogrammetry engine (such as ContextCapture), including command-line tools based on CloudCompare or a self-developed point cloud / model difference analysis algorithm library to realize 3D model generation and incremental calculation;

[0037] Voxelization Engine Service: An independent microservice that receives irregular incremental models (such as OBJ or PLY format) and outputs structured voxel datasets (such as GeoJSON or custom binary format).

[0038] Spatiotemporal correlation and injection service: The core module of the system, which maintains the "spatiotemporal anchor point" message queue, consumes anchor points and generates spatial query conditions, calls the spatial database query interface to execute targeted queries, and merges and writes the business information associated with the anchor points into the voxel ID list;

[0039] Database services include PostgreSQL + PostGIS (stores business data such as user, vehicle, and hazardous waste manifests, as well as volume data tables with spatial indexes) and MinIO / S3 (stores large unstructured files such as raw images, 3D model files, and on-site photos).

[0040] Visual Interaction Unit: A front-end web application for user interaction with the system, developed using modern front-end frameworks such as Vue.js or React.js, and integrated with a CesiumJS or Three.js 3D rendering engine (natively supporting 3DTiles format, efficiently loading high-precision 3D geospatial models, and providing GIS functions such as coordinate transformation and measurement analysis); it communicates with the central processing unit through RESTful API or GraphQL to realize data CRUD operations and visualization presentation, and supports interactive operations such as user clicking on the model to query waste information.

[0041] Technical effect

[0042] This invention achieves a technological leap from fuzzy "batch-region" tracing to precise "batch-location" tracing, completely breaking the "information black box." By clicking on any location in the digital model, users can instantly retrieve information such as the batch and origin of buried waste. It transforms unstructured spatial data into structured, database-manageable "volumetric information units," laying the foundation for spatial queries, statistical analysis, and precise correlation with business attributes, supporting upper-level intelligent applications. The tracing accuracy is significantly improved, with the economy version reaching 5-10 meters (mobile phone GPS accuracy) and the high-precision version reaching sub-meter accuracy (<1 meter, RTK). With centimeter-level technology, spatial resolution is improved by over 100 times; traceability efficiency is significantly improved, with traditional manual surveys taking several days to weeks, while the economic version of this invention reduces the time to within 5 seconds, and the high-precision version to within 3 seconds, resulting in an efficiency improvement of over 15,000 times; the accuracy of warehouse capacity calculation is improved, with traditional estimation errors exceeding 10%, while the economic version of this invention has an error of <5%, and the high-precision version has an error of <2%, resulting in a 5-fold improvement in measurement accuracy; the high error rate of traditional manual data entry is eliminated, with the economic version achieving an extremely low error rate through automatic system association, and the high-precision version achieving a near-zero error operation loop through fully automatic intelligent analysis.

[0043] The system-generated "spatiotemporal anchor points" (waste batch ID + high-precision spatiotemporal coordinates + on-site photos / sensor data) constitute an immutable digital evidence chain, reducing the difficulty and cost of investigation and evidence collection in pollution accident liability determination scenarios. It offers two implementation paths: a high-precision version and an economy version, adaptable to clients with different budgets (e.g., large landfills use the high-precision version, while small and medium-sized enterprises use the "ordinary drone + mobile app" economy version), reducing enterprise deployment costs. The core technology can be extended to large-scale mine spoil heaps, coal / ash yards at thermal power plants, and ore / bulk cargo storage yards at ports and wharves, providing accurate inventory calculation and operation management, creating additional economic value. It supports diversified business models such as Software as a Service (SaaS), project-based solutions, and data services (regular surveying + report delivery), bringing continuous revenue to technology transfer parties. To meet increasingly stringent environmental regulations, this initiative will help hazardous waste landfills achieve "full-process traceability and full-stage monitoring," reduce environmental pollution risks, and protect the ecological environment. It will also provide technical support for environmental pollution prevention and control and accountability, clarify the responsible parties for accidents, and safeguard public interests. Furthermore, it will promote the digital and intelligent upgrading of the hazardous waste treatment industry, guide the industry's technological development, and foster high-quality development of the environmental protection industry.

[0044] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating a preferred embodiment of the hazardous waste landfill traceability method based on spatiotemporal anchor points and volumetric models of the present invention.

[0046] In the diagram: 1 - Waste dumping, 2 - Spatial data acquisition (drone flight to acquire images), 3 - Spatiotemporal anchor point capture (vehicle-mounted equipment generates anchor point data packets), 4 - Data voxelization (incremental model is divided into voxels), 5 - Spatiotemporal anchor point and voxel association (business information is injected into voxels), 6 - Precise traceability and query (front-end visual interaction). Detailed Implementation

[0047] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0048] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of some components has been appropriately exaggerated in the drawings.

[0049] Example 1: A detailed method for precise traceability of hazardous waste landfills

[0050] This embodiment describes in detail and step by step the complete method flow for implementing the present invention.

[0051] Step 1: System initialization and data pre-binding to establish a baseline 3D model:

[0052] Before the operation began, drones were used to conduct standardized flight paths over the landfill units, collecting image data containing high-precision position and attitude information. This image data was used to generate a 3D benchmark model of the landfill area. We used Structure from Motion (SfM) and Multi-View Stereo (MVS) algorithms to accurately reconstruct the 3D information of the site. The 3D models generated by the SfM and MVS algorithms have high accuracy and can be compared at different time points, facilitating subsequent difference analysis.

[0053] Principles and formulas of 3D reconstruction:

[0054] Structure of Motion (SfM): The SfM algorithm reconstructs a 3D point cloud by calculating image points of the same object from different viewpoints, estimating the camera's position and pose. In 3D reconstruction, we first assume a linear relationship between each image point p and a 3D point P=(X,Y,Z) in world space: p=K[R|t]P. Here, p is a 2D image point, P is a 3D point, R and t are the camera's rotation matrix and translation vector, respectively, and K is the camera's intrinsic parameter matrix. By acquiring image data from multiple viewpoints, we can reconstruct the 3D point cloud by matching image points and optimizing these parameters.

[0055] Multi-view stereo matching (MVS): Building upon SfM, MVS further refines point cloud data by calculating pixel matching between images, increasing point cloud density and correcting geometric errors. By optimizing the matching relationship of each pixel, a more accurate 3D model is constructed.

[0056] Step 2: Capture the spatiotemporal anchor point of the tipping event

[0057] When a transport vehicle arrives at the landfill site and begins dumping, the system captures a "spatiotemporal anchor" data packet. This packet records the vehicle's position (X, Y, Z), a timestamp, and an event identifier. The spatiotemporal anchor accurately identifies the location and time of hazardous waste dumping and is crucial data for the system's traceability process. The spatiotemporal anchor generation formula is as follows: When a triggering event occurs, the system obtains precise three-dimensional coordinates (X, Y, Z) through the vehicle-mounted RTK-GNSS device and generates a spatiotemporal anchor data packet by combining sensor data or manual confirmation of the triggering event. The structure of this data packet is as follows: AnchorPoint={"batch_id": "HW-2025-A001", "timestamp": 1750898826123, "coordinate": (X, Y, Z), "source_type": "RTK_AUTO"}. Here, X, Y, and Z are the precise three-dimensional coordinates at the time of dumping, and the timestamp ensures the accuracy of the time, facilitating subsequent analysis and traceability.

[0058] The captured spatiotemporal anchor point data is transmitted to the backend platform in real time via wireless network and stored in the database for subsequent processing and querying.

[0059] Step 3: Obtain spatial data after the job is completed

[0060] After the landfill operation was completed, operators used drones to fly again, collecting image data and generating a second-phase 3D model (M2). This process ensured complete monitoring of the landfill area and provided data for subsequent change analysis.

[0061] By comparing the two 3D models (M1 and M2), we can calculate the newly added landfill model (incremental model) and obtain the 3D coordinates of the incremental area through geometric difference analysis. This process can be performed using the following formula for difference calculation:

[0062] ΔV=∑ i=

[0063] Where M2(Xi,Yi,Zi) and M1(Xi,Yi,Zi) represent the three-dimensional coordinate values ​​at time points T2 and T1, respectively. The incremental model ΔV represents the newly added landfill area.

[0064] Step 4: Spatial Variation Calculation and Structured Volumetric Subdivision

[0065] To facilitate the storage and retrieval of spatial data, the incremental model (ΔV) needs to be subdivided into smaller three-dimensional units (volumetric cells). These voxels are typically cubes, with side lengths that can be set as needed (e.g., 0.1 meters, 0.5 meters, etc.). Each voxel has a unique spatial index and additional attributes for easy subsequent querying and management.

[0066] The voxelization formula is: Vi = Voxelize(ΔV). Where Vi represents the voxelized element and ΔV is the incremental model.

[0067] Each voxel is assigned a unique three-dimensional coordinate index (a, b, c), which makes data storage structured and facilitates subsequent retrieval and association.

[0068] To efficiently store and query body metadata, the system employs two space-optimized indexing algorithms: octree and kd-dimensional tree, which optimizes the efficiency of spatial data access.

[0069] Octree Algorithm: An octree recursively divides a three-dimensional space into eight sub-regions, each containing a certain number of voxels. During a query, only the relevant sub-region needs to be accessed, greatly reducing computational complexity. Each leaf node contains voxel data, and each internal node represents a spatial region.

[0070] Octree construction formula: Node(a,b,c)=Octree(Vi)

[0071] KD-tree algorithm: A KD-tree is a binary tree structure generated by partitioning space according to its dimensions. Each node represents a spatial region, and queries are performed by traversing the tree structure to find voxels. Due to the efficiency of KD-trees in multidimensional data queries, they can quickly filter out voxels that match the specified query conditions.

[0072] kd-tree formula: Nodi = KDTree(Vi, split_dimension)

[0073] Step 5: Targeted Attribute Injection Based on Spatiotemporal Anchors

[0074] The system generates a 3D spatial query volume using the coordinates of a spatiotemporal anchor point. This volume can be shaped like a sphere, cylinder, etc., and its size is set according to actual needs, for example, a sphere with a radius of 5 meters. This query volume allows for efficient retrieval of all elements related to the spatiotemporal anchor point. Once a relevant element is found, the system injects all business information related to that anchor point (such as hazardous waste batches, transport vehicle information, etc.) into the attribute fields of that element, achieving a link between the spatiotemporal anchor point and spatial data. Through this process, different types of data (such as task information and spatial information) can be efficiently bound, facilitating traceability and querying.

[0075] Attribute injection formula: Vi = InjectAttributes(Vi, business_info)

[0076] Where Vi is the retrieved element, and business_info is the business information associated with the spatiotemporal anchor point.

[0077] Step Six: Data Persistence and Visualization Applications

[0078] All voxels with injected attributes will be stored in a spatial database, and a query index will be built to support fast spatial data retrieval. For user queries, the system communicates between the front-end and back-end via a RESTful API or GraphQL. Through the front-end visualization module, users can click on any location in the 3D model to query waste information at that location. The system will immediately retrieve the relevant voxel attributes from the database and display detailed waste information. This query process is completed through interaction with the back-end service.

[0079] Example 2: A system for precise traceability of hazardous waste landfills

[0080] This embodiment describes in detail a physical system capable of implementing the above method, which consists of cooperating hardware units and software modules.

[0081] 1. System Overall Architecture

[0082] This system is a typical IoT and digital twin convergence system, architecturally divided into a perception layer, a network layer, a platform layer, and an application layer. Physically, it mainly includes: an onboard perception unit, a spatial mapping unit, a central processing and data management unit, and a visualization and interaction unit.

[0083] 2. Vehicle-mounted sensing unit

[0084] This unit is the hardware generator for the "spatiotemporal anchor point," installed on a hazardous waste transport vehicle. Its core components include:

[0085] Positioning module: An industrial-grade high-precision RTK receiver that provides real-time, centimeter-level three-dimensional coordinates.

[0086] Trigger module:

[0087] Tilt sensor: A MEMS (Micro-Electro-Mechanical Systems) tilt sensor, firmly mounted on the truck bed frame, communicates with the main controller via CAN bus or RS-485 interface, with an accuracy better than 0.1 degrees.

[0088] (or) Physical button: A rugged, durable, industrial-grade physical button installed in a location easily accessible to the driver, providing a manual trigger signal.

[0089] Main Controller (MCU): An embedded microcontroller, such as an ARM Cortex-M series microcontroller. It is the brain of the vehicle unit and is responsible for:

[0090] Real-time analysis of NMEA-0183 format data output by the positioning module.

[0091] Listen for signals from the trigger module (angle change or button press).

[0092] Upon receiving the trigger signal, the coordinates and timestamp are "frozen" and packaged.

[0093] Obtain the preset vehicle / task ID from the communication module.

[0094] Communication module: A full-network compatible 4G / 5G Cat-1 / Cat-4 communication module, responsible for establishing long or short connections with the central processing unit via MQTT or HTTP / S protocols to ensure the reliable uploading of "spatiotemporal anchor" data packets.

[0095] 3. Spatial mapping unit

[0096] This unit is responsible for acquiring high-precision 3D scene data, and its core components include:

[0097] Flight platform: An industrial-grade multi-rotor UAV with RTK / PPK differential positioning capability, capable of achieving centimeter-level flight path accuracy and control point accuracy.

[0098] Mission payload: A full-frame, high-resolution (e.g., 45 megapixels or higher) orthophoto or tilt camera equipped with a low-distortion prime lens.

[0099] 4. Central Processing and Data Management Unit (the core of this system)

[0100] This unit is a complete backend software system deployed on a server (local or cloud), consisting of multiple decoupled microservice modules:

[0101] Data Access and Gateway Service: As the system entry point, it is responsible for handling MQTT connections and data reporting from the vehicle-mounted sensing unit, as well as uploading image data from the spatial mapping unit.

[0102] 3D Reconstruction and Spatial Analysis Services:

[0103] Embedded commercial or open-source photogrammetry engines (such as ContextCapture).

[0104] It includes a point cloud / model difference analysis algorithm library specifically optimized for this invention (such as a command-line tool based on CloudCompare or a self-developed algorithm).

[0105] The voxelization engine service is an independent microservice that receives irregular incremental models (such as OBJ or PLY formats) and outputs structured voxel datasets (such as GeoJSON or custom binary formats). This service is the technical carrier for realizing the innovative point of "structured partitioning" in this invention.

[0106] Spatiotemporal correlation and injection service (the "invention point" core module of the system): This service is the core of the technical solution of this system, and it is responsible for:

[0107] Maintain a pending "spatiotemporal anchor" message queue.

[0108] Anchor points in the consumption queue generate spatial query conditions based on their coordinates and preset rules (such as radius and precision source).

[0109] Call the query interface of the spatial database to perform a targeted query.

[0110] The query results (list of element IDs) are then merged and written with the business information associated with the anchor points.

[0111] Database services:

[0112] PostgreSQL + PostGIS: Used to store business data such as users, vehicles, and hazardous waste manifests, as well as the core entity data table with spatial indexes.

[0113] MinIO / S3: Object storage service used to store large, unstructured files such as raw images, 3D model files, and site photos.

[0114] 5. Visual Interactive Unit

[0115] This unit is the front-end interface for users to interact with this system; it is a web-based application.

[0116] Front-end framework: Developed using modern front-end frameworks such as Vue.js or React.js, responsible for UI components, business logic, and user interaction.

[0117] 3D rendering engine: CesiumJS or Three.js is used. CesiumJS is particularly suitable because it natively supports the 3DTiles format, can efficiently load and render large-scale, high-precision 3D geospatial models, and provides rich GIS functions such as coordinate transformation and measurement analysis.

[0118] API Client: Communicates with various microservices of the central processing unit via RESTful API or GraphQL to perform CRUD operations and visualize data. When a user clicks on a point in a 3D model in their browser, the front-end sends the coordinates of that point to the back-end. The back-end performs a spatial query and returns the associated waste information for that point. The front-end then renders this information onto the interface, completing a full "source-by-image" interaction.

[0119] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for tracing hazardous waste landfills based on spatiotemporal anchor points and volumetric models, characterized in that, Includes the following steps: (1) System initialization and data pre-binding to establish a benchmark three-dimensional model: Before the operation, the UAV flew along a standardized route to collect image data containing high-precision position and attitude information, and used the structure of motion restoration (SfM) and multi-view stereo matching (MVS) algorithms to generate a three-dimensional benchmark model M1 of the landfill area; (2) Capture the spatiotemporal anchor point of the dumping event: When the transport vehicle is dumping, the system captures the "spatiotemporal anchor point" data packet containing the waste batch ID, three-dimensional coordinates (X,Y,Z), timestamp, and trigger source through the on-board equipment, and transmits it to the back-end platform in real time for storage; (3) Acquire spatial data after the operation and calculate the incremental model: After the operation is completed, the UAV flies again to generate a three-dimensional model M2. Compare M1 and M2, and calculate the incremental model using ΔV=∑i= Calculate the incremental model ΔV of the newly added landfill body; M2(Xi,Yi,Zi) and M1(Xi,Yi,Zi) represent the three-dimensional coordinate values ​​at time points T2 and T1, respectively; (4) Structured voxel subdivision: According to the set side length, the incremental model ΔV is subdivided into voxel Vi with unique three-dimensional coordinate index (a,b,c) by Vi=Voxelize (ΔV), and the data access efficiency is optimized by using octree and kd tree. (5) Targeted attribute injection: Generate a 3D query body based on the spatiotemporal anchor coordinates, retrieve relevant elements, and inject business information into the attribute fields of the elements through Vi=InjectAttributes (Vi,business_info); business_info is the business information associated with the spatiotemporal anchor. (6) Data persistence and visualization query: Store the volume elements of related business information in the spatial database and establish an index. Users can click on any position of the model through the front-end visualization module to query waste information in real time; In step (4), the octree recursively divides the three-dimensional space into eight sub-regions according to Node (i,j,k)=Octree (Vi), and the kd tree generates a binary tree structure by dividing the dimension according to Nodei=KDTree (Vi,split_dimension). Both are used to reduce the computational amount of volume element query. In step (5), the three-dimensional query volume is a sphere or cylinder, and the business information includes hazardous waste batch number, source unit, waste type, and transport vehicle ID; the injected attribute fields are bound to the unique three-dimensional coordinate index (a,b,c) of the volume element to achieve a one-to-one correspondence between "spatiotemporal anchor point - volume element - business information".

2. The method according to claim 1, characterized in that, In step (1), the SfM algorithm satisfies the linear relationship p=K [R|t] P, where p is a two-dimensional image point, P is a three-dimensional spatial point, R and t are the camera's rotation matrix and translation vector, respectively, and K is the camera's intrinsic parameter matrix; the MVS algorithm refines the point cloud data through pixel matching, thereby improving the accuracy of the three-dimensional reference model M1.

3. The method according to claim 1, characterized in that, In step (2), the vehicle-mounted equipment includes an RTK-GNSS positioning module, a triggering module, a main controller, and a communication module; the spatiotemporal anchor data packet structure is AnchorPoint={"batch_id": "HW-2025-A001", "timestamp": 1750898826123, "coordinate": (X, Y,Z), "source_type": "RTK_AUTO"}.

4. A hazardous waste landfill traceability system based on spatiotemporal anchor points and a volumetric model, based on the method of any one of claims 1-3, characterized in that, This system is an IoT and digital twin fusion system, comprising an onboard sensing unit, a spatial mapping unit, a central processing and data management unit, and a visualization interaction unit. These units are connected via network layer communication. The onboard sensing unit generates spatiotemporal anchor point data packets; the spatial mapping unit collects image data and generates 3D models; the central processing and data management unit processes data, generates incremental models, decomposes voxels, and associates business information; and the visualization interaction unit provides 3D model display and query interaction functions.

5. The system according to claim 4, characterized in that, The vehicle-mounted sensing unit includes: a positioning module: an industrial-grade high-precision RTK receiver, providing three-dimensional coordinates with centimeter-level error; a triggering module: a MEMS tilt sensor or an industrial-grade physical button; a main controller: an embedded microcontroller, which parses positioning data, processes trigger signals, and packages anchor point data; and a communication module: a full-network compatible 4G / 5G Cat-1 / Cat-4 module, which uploads data via MQTT or HTTP / S protocols.

6. The system according to claim 4, characterized in that, The space mapping unit includes: a flight platform: an industrial-grade multi-rotor UAV with RTK / PPK differential positioning capabilities, achieving centimeter-level flight path accuracy; and a mission payload: a full-frame, ≥45-megapixel orthophoto or oblique photography camera equipped with a low-distortion fixed-focus lens.

7. The system according to claim 4, characterized in that, The central processing and data management unit includes multiple microservice modules: data access and gateway service: processing data reporting from vehicle-mounted units and image uploading from surveying and mapping units; 3D reconstruction and spatial analysis service: integrating a photogrammetry engine to calculate incremental models; voxelization engine service: dividing incremental models into structured voxels; spatiotemporal association and injection service: associating spatiotemporal anchors with voxels and injecting business information. Database services: PostgreSQL + PostGIS for storing business and data metadata, and MinIO / S3 for storing large unstructured files.

8. The system according to claim 4, characterized in that, The visualization interaction unit is developed using the Vue.js or React.js front-end framework and integrates the CesiumJS or Three.js 3D rendering engine. It communicates with the central processing unit through RESTful API or GraphQL, supports model rotation, scaling, and click-to-query, and the query results can be exported to Excel or PDF format.

Images