Multi-communication technology fusion unmanned vehicle digital twin positioning system based on STM32

Abstract

The invention discloses an STM32-based multi-communication technology fusion unmanned vehicle digital twin positioning system, and aims to solve the problems of insufficient positioning precision and difficulty in multi-communication technology integration caused by a multi-path effect and non-line-of-sight propagation in a complex industrial environment. The system adopts a hardware layer double-board modular design, integrates UWB, GPS / BD dual-mode satellites, IMU (MPU6050), electronic compass (QMC5883), WiFi / BT and NB-IoT modules, guarantees signal integrity and electrical isolation through PCB optimization and multi-stage LDO voltage stabilization, and constructs a multi-source sensor fusion hardware basis. According to the system, a unique digital ID of equipment is established based on an industrial internet identification analysis system to realize cross-domain secure transmission of data, an electromagnetic path is rehearsed by using a ray tracing technology based on a digital twin closed-loop optimization mechanism, multi-domain characteristic differences of physical and virtual signals are extracted through wavelet transform, and optimal position estimation is output in combination with Bayesian inference. And meanwhile, an A * / D * Lite algorithm is adopted to realize dynamic path planning and obstacle avoidance. A four-layer architecture including a hardware layer, a network layer, a data layer and an application layer is integrally constructed, and real-time interaction of a physical entity and a virtual model and iterative optimization of a positioning result are realized. The system improves the positioning precision of complex scenes, effectively inhibits multipath interference, improves the reuse rate of hardware resources, and is suitable for the high-precision positioning requirements of unmanned vehicles in the fields of intelligent logistics, industrial automation and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of unmanned vehicle positioning technology, in particular to an unmanned vehicle positioning system design method based on multi-communication technology and digital twin technology, a multi-communication technology integrated development board design method, and a digital twin system design method. It is suitable for complex environment positioning fields such as intelligent logistics and industrial automation. BACKGROUND

[0002] Under the wave of the rapid development of intelligent driving and industrial Internet of Things, unmanned equipment is accelerating the transformation to high-precision autonomous operation. In complex industrial scenarios, accurate positioning technology has become the core support for the stable operation of unmanned equipment. Although the global satellite navigation system has created significant value in many fields due to its high precision and wide coverage, its signal is easily blocked by buildings, and its positioning function is basically ineffective in indoor scenarios, making it difficult to meet the stringent requirements of industrial environment coverage.

[0003] To solve the positioning needs of indoor and outdoor scenarios, wireless positioning technologies such as Wi-Fi, Bluetooth, ultra-wideband (UWB), and 5G cellular networks have emerged and quickly spread. Wi-Fi positioning is based on ranging intersection or fingerprint matching principles, relying on widely laid network facilities and low-cost advantages, and is widely used in indoor scenarios, but signal strength fluctuations and multipath effects limit its positioning accuracy. Bluetooth positioning uses ultra-high frequency radio waves, with a coverage range of 10-100 meters, low power consumption, and high-speed transmission characteristics, and performs outstandingly in short-range positioning, but also faces the problem of signal interference and reduced accuracy in complex environments. UWB technology, as a carrierless communication scheme, has the advantages of high transmission rate, low transmission power, and strong penetration, but it is highly dependent on line-of-sight propagation, and once obstructed by obstacles, the ranging accuracy will be severely reduced. 5G positioning uses uplink time difference measurement and beamforming reference signals to solve the problem, and theoretically can achieve a positioning accuracy of 0.3 meters in the millimeter wave band, but non-line-of-sight path loss, insufficient base station deployment density, and high cost and difficulty in popularization limit its large-scale practical application.

[0004] These single indoor positioning technologies have certain applications in specific scenarios, but in complex environments such as industrial plants and warehouse centers, signal multipath effects and non-line-of-sight propagation interference make it difficult for their positioning accuracy to meet the stringent standards of industrial production. Current researches mostly use multi-sensor fusion architecture to integrate the advantages of different positioning technologies, achieve coverage range expansion, positioning accuracy optimization, and interference resistance improvement, and Kalman filtering and its derivative algorithms are the most common for time and space registration and error compensation. However, these researches are mostly limited to algorithm-level optimization, and in actual engineering applications, they have exposed problems such as low hardware integration, difficulty in adapting to multiple communication protocols, and significant fluctuations in positioning accuracy when deploying across scenarios, which seriously affects the reliability of real-time decision-making by unmanned equipment.

[0005] In summary, the prior art solution cannot meet the actual industrial production requirements in terms of positioning accuracy due to the multi-path and non-line-of-sight problems of signal propagation in complex industrial scenarios, and the current positioning method integrated with multiple communication technologies lacks a practical hardware and software collaborative implementation framework. The present application proposes a multi-communication technology integrated unmanned vehicle digital twin positioning system based on STM32 to solve the above problems. Based on the multi-source fusion positioning embedded system of the STM32H7 series high-performance microcontroller, the optimized circuit design can realize the parallel access and real-time data processing of UWB, Wi-Fi, Bluetooth and 5G multi-mode communication modules. Equipped with a channel digital twin platform with real-time interaction capability, it integrates the core functions of data management, high-precision positioning and intelligent path planning. The unmanned vehicle digital twin positioning system includes a four-layer framework of hardware layer, network layer, data layer and application layer, realizing real-time interaction between physical entities and virtual models. SUMMARY

[0006] The present application proposes an unmanned vehicle positioning system based on multi-technology fusion and digital twinning, comprising the following steps:

[0007] Step one, modular design of unmanned vehicle fusion positioning embedded system

[0008] The present application realizes the modular architecture of the unmanned vehicle fusion positioning embedded development board with ARM-M4 microcontroller as the core, and builds a double-board interactive hardware platform composed of a drive control board and a fusion positioning board. The drive control board uses STM32F407VET6 as the main control chip, carries HTD8236 motor drive chip, accurately controls four-way Hall encoder motor through PWM signal, and realizes high-precision angle control of ±0.5° by combining with the omni-directional movement characteristics of Mecanum wheel. At the same time, it integrates OpenMV vision module, transmits structured road information processed by algorithm through UART interface, and provides input data stream for unmanned vehicle navigation decision. The fusion positioning development board takes AS608-60QFN microcontroller as the core processing unit, is equipped with SPI / UART / I2C / GPIO multi-type interface resources, integrates inertial measurement unit (MPU6050), electronic compass (QMC5883), ultra-wideband transceiver module (UWB-BU01), GPS / BD dual-mode satellite positioning module, WiFi / BT composite communication module and NB-IoT cellular communication unit through PCB layout optimization, builds a multi-sensor data fusion hardware foundation, and realizes electrical connection optimization and signal integrity protection of each functional module through circuit design. The two boards establish a serial data channel through the SPI interface, and the drive control board receives the positioning information processed by the fusion positioning development board and executes motion control.

[0009] Step two, real-time interactive unmanned vehicle digital twin software system design

[0010] In view of the problem that the positioning accuracy is not high due to the multi-path and non-line-of-sight problems in complex industrial scenes, and the problem of realizing the flow and sharing of information in multiple fields, the application proposes a real-time interactive unmanned vehicle digital twin software system design combining industrial internet identification analysis and digital twin technology. Relying on the industrial internet identification analysis system, a unique digital ID is allocated to the unmanned vehicle, and the hardware layer data and space-time information are bound into standardized data flow and uploaded to the secondary node platform to realize cross-domain data safe transmission. Based on the three-dimensional engine, a twin environment is constructed, the electromagnetic propagation path is preformed through the ray tracing technology, and the time domain / frequency domain / Doppler domain features of physical signals and virtual signals are extracted through wavelet transform. The multi-domain waveform difference degree is calculated, the difference degree is taken as the input to construct the multi-base station joint likelihood function, and the posterior probability is deduced by combining the prior position distribution and the Bayesian inference. The posterior probability is maximized, and finally the optimal position estimation is output and mapped into the unmanned vehicle digital twin model to realize the closed-loop optimization of unmanned vehicle positioning. According to the positioning result, A* / D*Lite algorithm is used to realize global path planning and dynamic obstacle avoidance. The application realizes the core functions including data management, real-time positioning and dynamic path planning. Specifically, the data management module stores real-time data and historical operation logs in layers, supports offline data import in CSV format and error statistical analysis, and provides data support for algorithm optimization. The positioning module maps the vehicle position in real time through the 3D twin model, supports manual simulation positioning test and multi-algorithm dynamic switching, enables UWB+IMU tight coupling mode in indoor scenes, switches to GPS+SLAM loose coupling mode outdoors, and opens Kalman filter parameter configuration interface to adapt to different environments. The path planning module dynamically generates global and local paths combined with A* / D*Lite algorithm, and switches the planning mode in real time based on scene requirements.

[0011] Step three, multi-source heterogeneous unmanned vehicle positioning software and hardware integrated design based on digital twin

[0012] This invention achieves real-time interaction and closed-loop coordinate optimization between the physical entity and virtual model of an unmanned vehicle (UAV) through a four-layer closed-loop architecture comprising a hardware layer, a network layer, a data layer, and an application layer. The hardware layer integrates MPU6050, QMC5883, UWB-BU01, and GPS / BD dual-mode satellite multi-source sensor positioning modules to collect real-time motion attitude data, UWB signal strength, and satellite positioning information of the UAV. This UAV status data is then uploaded to the network layer via the ESP32 dual-frequency module. The network layer, relying on the Industrial Internet Identifier Resolution System, assigns a unique digital ID to the UAV, binding hardware layer data with spatiotemporal information into a standardized data stream uploaded to the secondary node platform, achieving unified encoding and secure transmission of cross-domain data. The data layer, based on the application layer's twin model, calculates the optimal position estimate using ray tracing technology to pre-simulate the electromagnetic propagation path and obtain channel characteristics, achieving closed-loop optimization of UAV positioning. The application layer maps the optimized positioning data to the virtual UAV's trajectory and then sends path correction commands to the hardware layer actuators via a reverse control channel, realizing path planning functionality.

[0013] Compared with the prior art, the advantages of the present invention are:

[0014] (1) This invention constructs a dual-board interactive hardware platform consisting of a drive control board and a fusion positioning board. The drive control board uses an STM32F407VET6 as the main control chip, controls four Hall-coded motors via PWM signals, and integrates an OpenMV vision module. The fusion positioning board uses an AS608-60QFN as the core processor and integrates an inertial measurement unit (MPU6050), an electronic compass (QMC5883), an ultra-wideband transceiver module (UWB-BU01), and a GPS / BD dual-mode satellite positioning module. Compared with traditional single-technology solutions, this invention realizes the hardware foundation for data acquisition using multiple communication technologies, providing physical support for high-precision positioning in complex environments.

[0015] (2) This invention proposes a real-time interactive digital twin software system design for unmanned vehicles, combining the Industrial Internet Identifier Resolution System and digital twin technology. The system relies on the Unity3D engine to build a twin visualization platform, achieving accurate mapping between physical and virtual spaces. Simultaneously, it integrates the Industrial Internet Identifier Resolution System to assign unique digital IDs to unmanned vehicles, enabling secure cross-domain transmission of device status and positioning data. Finally, through the integration of three core functions—data management, real-time positioning, and dynamic path planning—the upper-level software system forms a closed-loop control architecture that combines virtual and physical elements.

[0016] (3) To address the lack of a practical hardware-software collaborative implementation framework for positioning methods integrating multiple communication technologies, this invention proposes a four-level architecture: hardware layer, network layer, data layer, and application layer. The hardware layer adopts a dual-board modular design, optimizing PCB layout to achieve electrical isolation and signal integrity assurance for multiple source sensors and communication modules. The network layer relies on the Industrial Internet identifier resolution system to construct a "one vehicle, one code" full-domain data hub, enabling secure cross-platform transmission of equipment status and location information. The data layer designs a three-level processing pipeline, ensuring real-time performance through cascaded priority scheduling. The application layer develops a virtual-real interactive console based on the Unity3D engine, supporting electromagnetic interference pre-simulation to obtain multi-domain channel characteristics. This architecture solves the problems of poor real-time performance and high deployment costs caused by the hardware-software separation in traditional solutions through hardware resource isolation, hierarchical data flow optimization, and cross-layer instruction closed-loop. Attached Figure Description

[0017] Appendix Figure 1 This is the overall system framework diagram of the present invention;

[0018] Appendix Figure 2 This is a circuit schematic diagram of the main control chip of the drive control board of this invention;

[0019] Appendix Figure 3 This is a circuit diagram of the motor drive of the drive control board of this invention;

[0020] Appendix Figure 4 This is the circuit schematic diagram of the integrated positioning development board main control chip of the present invention;

[0021] Appendix Figure 5 This is the circuit schematic diagram of the GPS / BD dual-mode satellite integrated positioning development board of this invention;

[0022] Appendix Figure 6 This is a circuit schematic diagram of the ultra-wideband transceiver module of the integrated positioning development board of this invention;

[0023] Appendix Figure 7 This is a circuit schematic diagram of the WiFi / BT composite communication module of the integrated positioning development board of the present invention;

[0024] Appendix Figure 8 This is a circuit schematic diagram of the NB-IoT cellular communication unit of the integrated positioning development board of this invention;

[0025] Appendix Figure 9 This is an interface diagram of the positioning function module of the software system of the present invention. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0027] Step 1: Modular Design of Unmanned Vehicle Fusion Positioning Embedded System

[0028] This invention designs a modular architecture for an embedded development board for fusion positioning of unmanned vehicles, based on an ARM-M4 microcontroller, and constructs a dual-board interactive hardware platform consisting of a drive control board and a fusion positioning development board. (See attached diagram) Figure 1 The hardware layer's drive control board uses the STM32F407VET6 main control chip, comprising a motion execution module and a vision module. The motion execution module is equipped with an HTD8236 motor driver chip, precisely controlling four Hall-coded motors via PWM signals and combining this with the omnidirectional movement characteristics of the Mecanum wheels to achieve high-precision angle control of ±0.5°. The vision module transmits structured road information processed by algorithms through a UART interface, providing input data streams for autonomous vehicle navigation decisions. The fusion positioning development board uses the AS608-60QFN microcontroller as its core processing unit, equipped with multiple interface resources including SPI / UART / I2C / GPIO. Through PCB layout optimization, it integrates a positioning module consisting of an ultra-wideband transceiver module (UWB) and GPS / BD dual-mode satellites, an auxiliary positioning module consisting of an inertial measurement unit (MPU6050), an electronic compass (QMC5883), and ultrasonic sensors, and a composite communication module consisting of WiFi / BT and NB-IoT, constructing the hardware foundation for multi-sensor data fusion. Circuit design optimizes the electrical connections of each functional module and ensures signal integrity. A serial data channel is established between the two boards via an SPI interface. The drive control board receives the positioning information processed by the fusion positioning development board and executes motion control.

[0029] As attached Figure 2 The driver control board comprises a core system consisting of chip circuitry, clock circuitry, power supply circuitry, reset circuitry, and general-purpose input / output circuitry. Through hardware design and software programming, this circuitry enables functions such as general-purpose I / O port control, multi-source data acquisition, serial port control, and I2C bus storage. (See attached image.) Figure 3 This is the circuit schematic of the motor drive in the motion execution module. The circuit controls the switching state of the drive module through the PB10_TX3 signal, and adjusts the drive current intensity by combining the resistor network to achieve precise control of the speed or direction of motion of the multi-channel motor.

[0030] Appendix Figure 4 This is the circuit schematic of the AS608-60QFN main chip for the integrated positioning development board. Power management uses a three-stage LDO regulator to achieve progressive voltage domain control of +3.3V / +1.8V / +1.2V, with each stage output equipped with a 4.7μF filter capacitor to suppress ripple. The communication interface integrates a hybrid bus structure, I... 2 The C bus supports the connection of magnetometers and inertial sensors, receives satellite positioning data via asynchronous serial port, connects to the UWB positioning unit via SPI bus, and configures RC filter networks for key signal paths to effectively suppress high-frequency interference of ≥20MHz.

[0031] The fusion positioning development board of this invention integrates an inertial measurement unit (MPU6050), an electronic compass (QMC5883), an ultra-wideband transceiver module (UWB-BU01), and a GPS / BD dual-mode satellite positioning module. The inertial measurement unit (MPU6050) and the electronic compass (QMC5883) are connected to the fusion positioning board. (See attached diagram.) Figure 5 The GPS / BD dual-mode satellite positioning module uses the CPQ8 RF front-end chip to process 1575MHz (GPS L1) and 1561MHz (BeiDou B1) dual-band satellite signals. It optimizes the stability of the 3.3V power supply with a 0.1μF decoupling capacitor and achieves low-power mode switching with the GPS_EN enable pin. (See attached image) Figure 6 The UWB positioning unit interacts with the main control chip at high speed via the SPI bus. The EXTON / RESET control signal is configured with an RC filter network to eliminate electromagnetic interference. The three-stage LDO architecture outputs core voltages of 3.3V / 1.8V / 1.2V respectively, and each stage output is equipped with a 4.7μF filter capacitor bank to ensure power purity.

[0032] The communication module designed in this invention adopts a dual-mode architecture of WiFi / BT communication unit and NB-IoT communication unit. (See attached diagram) Figure 7 As shown, the WiFi / BT communication unit is built on the ESP32-C3 chip. A 0.1μF ceramic capacitor C50 is connected in parallel to its VDD33 power supply pin to suppress high-frequency ripple, while multiple GND pins are used for grounding to optimize heat distribution. In the communication interface design, GPIO16_TXD and GPIO7_RXD constitute the main UART channel and support AT command interaction, while programmable pins (such as PD11_IO) implement deep sleep wake-up and mixed-signal functions. Further details are provided in the appendix. Figure 8 The NB-IoT communication unit shown uses the Quectel EC-01F module to achieve cellular IoT connectivity. Specifically, it connects to the ANT_NB antenna through a 50Ω impedance matching line to ensure the integrity of the radio frequency signal. The SIM card interface signal line is connected in series with a 4.7kΩ current-limiting resistor R31 to provide ESD protection, and a power storage capacitor is configured on the VBAT1 pin to cope with sudden current demands.

[0033] Step 2: Real-time interactive digital twin software system for unmanned vehicles

[0034] To facilitate information flow and sharing across multiple application scenarios involving regions, entities, industries, and platforms, the unmanned vehicle digital twin software system utilizes an Industrial Internet Identifier Resolution Secondary Node Platform to establish a unique digital identity mapping for all elements. This enables real-time status data of unmanned vehicles to achieve interconnectivity and twin collaboration based on a unified identifier. Upon vehicle startup, the software system registers a unique digital ID with the identifier resolution secondary node, achieving global binding of "one vehicle, one code." Motion attitude data, UWB signal strength, and satellite positioning information collected in real-time by multi-source sensors are encapsulated into timestamped binary data frames and uploaded to the Industrial Internet Identifier Resolution Secondary Node Data Platform via the ESP32 dual-frequency communication module. The uploaded data generates a structured data packet containing the device's unique identifier code according to a preset identifier template and is synchronized to the platform database through the identifier registration interface. During this process, the secondary node dynamically manages the identifier mapping rules to ensure cross-regional data interconnectivity. Operators and upper-level platforms in the industrial field can query the identifiers through external interfaces. The Industrial Internet Identifier Resolution Secondary Node Data Platform resolves the corresponding unmanned vehicle's location and status data under the identifier code, achieving a digital mapping of the physical world.

[0035] To address the insufficient positioning accuracy caused by multipath and non-line-of-sight interference in complex industrial scenarios, this invention implements a digital twin software system for unmanned vehicles (UAVs) that combines digital twin technology with a closed-loop optimization mechanism. First, spatial geometric parameters and surface texture data are acquired using augmented reality measurement technology to generate a high-precision 3D virtual environment model. Then, a twin visualization platform is built using the Unity3D engine, ensuring consistency between the virtual space and the physical scene through coordinate system alignment and physical parameter configuration. During operation, physical layer sensor data is parsed and used in real-time to drive the virtual UAV's trajectory mapping. Ray tracing technology is used to pre-simulate the electromagnetic propagation path, and wavelet transform is employed to extract time-domain, frequency-domain, and Doppler-domain features of physical and virtual signals. By comparing the differences between virtual base station signals and measured data from physical base stations, a posterior probability model is generated by fusing multi-base station data using Bayesian inference. Maximizing the posterior probability, the optimal position estimate is finally output and mapped into the UAV digital twin model, achieving closed-loop optimization of UAV positioning. Based on the positioning results, A* / D*Lite algorithms are used to implement global path planning and dynamic obstacle avoidance. Figure 9 This is the system's interface diagram, which fully implements the core functions including data management, real-time positioning, and dynamic path planning.

[0036] The data management module enables data control through real-time data management, historical data management, statistical analysis, and import / export functions. In real-time data management, the wireless signal data module collects and stores real-time data on the unmanned vehicle and base station, including Bluetooth, WiFi, RSSI values, time delay, and signal strength fluctuation curves, while simultaneously recording device power-on timestamps and communication link status. The positioning coordinate data module stores positioning data, accelerometer attitude angle data, and base station absolute coordinate information in time series, constructs a multi-source coordinate system transformation matrix, and generates positioning trajectory snapshots by time slice. Historical data management employs a hybrid storage architecture, storing structured logs in a relational database and unstructured trajectory data in a distributed file system, supporting fast timestamp-based retrieval and scene playback, forming a historical database that can be retrieved in different time periods. The statistical analysis function performs multi-dimensional analysis based on real-time and historical data. By comparing positioning results, it merges preliminary positioning data with base station coordinates to generate optimized positioning coordinates, simultaneously stores obstacle avoidance trigger records, supports offline retransmission of raw image data, and intermediate results of multi-source data fusion algorithms, supporting error comparison, positioning accuracy comparison, and multi-algorithm trajectory comparison. The import and export functions enable data migration and interaction. All data is indexed by the device's unique ID and timestamp, meeting the needs of segmented retrieval, backtracking, and multi-scenario management.

[0037] The positioning module integrates multi-positioning method management and positioning process processing capabilities, supporting real-time location display, manual simulated positioning, and switching and management of positioning algorithms. The 3D twin interface, based on the geometric and behavioral models of physical entities, renders the vehicle's pose in real-time using the Unity engine, achieving synchronized virtual and real-world display. The positioning algorithm interface includes a configurable Kalman filter, supporting dynamic adjustment of process noise and observation noise weights. It combines virtual and real data comparison to achieve self-optimization of positioning accuracy and enables switching and management of positioning algorithms. Furthermore, the module has a built-in digital twin positioning method interface. When a method is selected, it can simulate the vehicle's potential positioning location, constructing a twin system of electromagnetic signal propagation in the current environment. It supports calling CloudRT tools or dynamically presenting the signal propagation route through animation. After positioning is initiated, the host computer sends a command to trigger the slave computer to poll all base stations and activated unmanned vehicles, collecting signals in real-time and transmitting them back to the host computer. The host computer processes the data based on the selected positioning method and displays the positioning results and trajectory.

[0038] The path planning module integrates multi-scenario path planning management capabilities, supporting path display, planning simulation, and switching and management of planning algorithms. During the autonomous vehicle's operation, RSSI, latency, position, and obstacle information are fed back every second. The host computer replans the path accordingly, and if the path changes significantly, it resends instructions to the slave computer. Global path planning generates an initial path based on the selected planning algorithm, and maps the path data to a virtual model through a digital twin engine. Local obstacle avoidance calculates the dynamic trajectory in real time using an algorithm, and the obstacle avoidance process is displayed synchronously on the twin interface. Furthermore, it allows setting task scenarios such as the autonomous vehicle grabbing corresponding numbered and colored wooden blocks and returning to a specified location. After selecting a scenario, real-time positioning is triggered, and the positioning data is input into the selected path planning algorithm. The host computer generates a planned path instruction to drive the autonomous vehicle along the path. The module also has a planning simulation function, calculating the planned path distance, task completion time, and obstacle avoidance information through the twin path, and using a built-in evaluation algorithm to analyze the planned path and optimize it to find the optimal path. The planning algorithm switching interface offers multiple strategy switching options (such as RRT and TEB). By adjusting the cost function weights (path length, energy consumption), it optimizes decision preferences and, combined with the multi-vehicle priority scheduling mechanism of the collaborative operation configuration module, achieves conflict resolution and optimal resource allocation. The system as a whole follows a two-way driving principle: real-time data from physical entities drives the synchronous update of the virtual model, while the algorithm optimization results in the virtual space are fed back to the physical controller through low-latency communication, forming a closed-loop control.

[0039] Step 3: Multi-source heterogeneous unmanned vehicle positioning software and hardware integration architecture based on digital twin

[0040] As attached Figure 1 This invention achieves bidirectional real-time interaction between the physical entity and virtual model of an unmanned vehicle through a four-layer architecture: hardware layer, network layer, data layer, and application layer. The hardware layer is based on an ARM-M4 microcontroller to build the core processing unit. It employs a modular design to integrate a drive control development board and a multi-communication technology fusion positioning development board with drive control. It collects raw data from the MPU6050 / QMC5883 / GPS / UWB of the fusion positioning development board in real time, and encapsulates it into timestamped binary data frames after preprocessing. In the reverse communication link between the inertial measurement unit (MPU6050) and the electronic compass (QMC5883), motion control commands (including target pose and velocity vectors) issued by the application layer are transmitted to the network layer through a port. After being parsed by the ESP32 module, they are distributed to the motor controller via the hardware layer bus. Simultaneously, the hardware layer feeds back the command execution status to the application layer in real time through the same channel.

[0041] All data is uploaded to the network layer via the dual-frequency ESP32 communication module. The network layer, relying on the Industrial Internet Identifier Resolution Secondary Node Platform, encodes and binds the device identity with the spatiotemporal information collected by the hardware layer according to a preset identifier template, and synchronizes it to the platform database through the identifier registration interface. During this process, the secondary node platform dynamically manages the identifier mapping rules to ensure cross-regional data interconnection. Status data flows from the network layer to the data layer. Operators and upper-level platforms in the industrial field can query the identifiers through external interfaces, obtain the data storage address returned by the secondary node platform, and access the storage address to retrieve the corresponding unmanned vehicle status data information, achieving a digital mapping of the physical world. Control data flows from the application layer through the data layer to the network layer, updating historical data under the identifier encoding. Instructions are transmitted to the actual execution module via the identifier resolution platform to achieve optimized control of the physical object.

[0042] The data layer receives multi-source state data transmitted from the network layer via Industrial Internet identifier resolution, as well as electromagnetic propagation multi-domain features and algorithm control commands pre-simulated by the digital twin model in the application layer. By utilizing multi-domain features and calculating differences, a joint likelihood function of multiple base stations is constructed. This is then used to derive posterior probabilities through prior location distribution, iteratively optimizing the result to output the optimal positioning outcome. This result is fed back to the application layer, driving the virtual-real location mapping of the digital twin model and supporting the path planning algorithm in generating obstacle-avoiding paths that avoid multi-path interference zones. Simultaneously, it is transmitted back to the network layer to update identifier data, enabling the iterative evolution of the positioning model.

[0043] The application layer constructs a digital twin model of the autonomous vehicle using a 3D engine. Ray tracing technology is used to simulate the propagation path of electromagnetic signals in complex environments, extracting multi-domain channel features including time delay, angle, and Doppler shift. These channel features are standardized through the Industrial Internet Identifier Resolution System and transmitted to the data layer, providing prior environmental knowledge for the positioning algorithm. This iteratively optimizes the position of the node to be located, minimizing discrepancies and outputting the optimal position estimate. The optimized positioning result generated by the data layer is fed back to the application layer, which maps it to the motion trajectory of the virtual autonomous vehicle and generates a global path plan using algorithms such as A* / D*Lite. The optimized path correction command is transmitted back to the hardware layer actuators via the Industrial Internet Identifier Resolution Platform. Simultaneously, the motion data collected in real time by the hardware layer is transmitted back to the data layer through the network layer to update the channel model parameters.

Claims

1. A digital twin positioning system for unmanned vehicles based on STM32 and integrating multiple communication technologies, characterized in that, Includes the following steps: Step 1: Modular Design of Unmanned Vehicle Fusion Positioning Embedded System The fusion positioning embedded system adopts a dual-board modular design consisting of a drive control board and a fusion positioning circuit board. The drive control board uses the STM32F407VET6 as the main control chip, controls four Hall encoder motors through PWM signals, and integrates an OpenMV vision module. The fusion positioning circuit board uses AS608-60QFN as the core processor and integrates a multi-source sensor module and a multi-mode communication module. Step 2: Design of a Real-Time Interactive Digital Twin Software System for Autonomous Vehicles The unmanned vehicle digital twin software system's digital twin platform relies on the industrial internet identifier resolution system to establish a unique digital ID for the device, enabling secure cross-domain data transmission. Based on the digital twin closed-loop optimization mechanism, it uses ray tracing technology to pre-simulate the electromagnetic path, extracts the multi-domain feature differences between physical and virtual signals through wavelet transform, and outputs the optimal position estimate by combining Bayesian inference. At the same time, it uses the A* / D*Lite algorithm to realize dynamic path planning and obstacle avoidance. Step 3: Integrated Design of Multi-Source Heterogeneous Unmanned Vehicle Positioning Software and Hardware Based on Digital Twin The unmanned vehicle (UAV) positioning hardware and software integrates a four-layer closed-loop architecture: hardware layer, network layer, data layer, and application layer. This enables real-time interaction and closed-loop coordinate optimization between the physical entity of the UAV and its virtual model. The hardware layer collects the UAV's motion attitude data, UWB signal strength, and satellite positioning information in real time. The network layer, relying on the Industrial Internet Identifier Resolution System, assigns a unique digital ID to the UAV, binding hardware layer data with spatiotemporal information into a standardized data stream uploaded to the secondary node platform, achieving unified encoding and secure transmission of cross-domain data. The data layer optimizes the UAV positioning closed-loop based on the application layer's twin model. The application layer maps the optimized positioning data to the virtual UAV's trajectory and then sends path correction commands to the hardware layer's actuators via a reverse control channel, realizing path planning functionality.

2. The modular design of the unmanned vehicle fusion positioning embedded system according to claim 1, characterized in that, The fusion positioning circuit board includes: Main control chip: The core processing unit is the AS608-60QFN microcontroller, equipped with multiple interface resources such as SPI / UART / I2C / GPIO; Positioning module: The ultra-wideband transceiver module (UWB-BU01) is connected to the main control chip via the SPI bus, and the GPS / BD dual-mode satellite module is configured with a CPQ8 RF front-end chip to process dual-frequency signals; Auxiliary positioning module: The inertial measurement unit (MPU6050) and the electronic compass (QMC5883) are connected via I... 2 C-bus access; Communication module: The WiFi / BT composite communication unit is built on the ESP32-C3 chip, and the NB-IoT cellular unit adopts the Quectel EC-01F module.

3. The embedded system according to claim 2, characterized in that, The circuit design of the UWB positioning unit includes: The UWB positioning unit interacts with the main control chip at high speed via the SPI bus, and the EXTON / RESET control signal is configured with an RC filter network to eliminate electromagnetic interference. The three-stage LDO architecture outputs core voltages of 3.3V / 1.8V / 1.2V respectively, and each stage output is equipped with a 4.7μF filter capacitor bank to ensure power purity.

4. The embedded system according to claim 2, characterized in that, The circuit design of the communication module includes: A 0.1μF ceramic capacitor is connected in parallel to the VDD33 power supply pin in the WiFi / BT unit to suppress high-frequency ripple; In the NB-IoT unit, a 4.7kΩ current-limiting resistor is connected in series with the SIM card interface signal line to provide ESD protection, and an energy storage capacitor is configured on the VBAT1 pin to deal with sudden current.

5. The design of the real-time interactive unmanned vehicle digital twin software system according to claim 1, characterized in that, The design of the software system includes: The Industrial Internet Identifier Resolution Second-Level Node Platform is used to bind a unique digital ID to the unmanned vehicle and encapsulate it into a binary data frame with a timestamp. Spatial geometric parameters and surface texture data are acquired using augmented reality measurement technology. A twin visualization platform is built using the Unity3D engine. The electromagnetic propagation path is pre-simulated using ray tracing technology, and the non-line-of-sight propagation path is reconstructed using ray tracing technology. Wavelet transform is used to extract time-domain, frequency-domain, and Doppler-domain features of physical and virtual signals. A joint likelihood function for multiple base stations is constructed using multi-domain waveform differences as input. A posterior probability model is derived through Bayesian inference, and the optimal location estimate is output by maximizing the posterior probability.

6. The integrated hardware and software design for multi-source heterogeneous unmanned vehicle positioning based on digital twins according to claim 1, characterized in that, The operating logic of the hardware and software architecture includes: Hardware layer: Collects data from multiple sensor sources and uploads it to the network layer via the ESP32 dual-frequency module; Network layer: By binding device identity with spatiotemporal information codes through the Industrial Internet Identifier Resolution System, secure cross-domain transmission is achieved; Data layer: Calculates the difference degree based on multi-domain features of the channel, and iteratively optimizes the posterior probability to generate the positioning result; Application layer: The optimized positioning results are mapped to a virtual trajectory, and path correction instructions are sent to the hardware actuators via the reverse control channel.

Images