Keywords

1 Introduction

With the rapid development of technology, digital and intelligent technologies have demonstrated strong potential and broad prospects across various domains. In the field of transportation, especially in driving training and vehicle management, the application of digital technology is gradually transforming traditional driving school vehicle systems. As one of the core technologies of digital transformation, Digital Twin technology boasts prominent advantages in simulating the digital image, operational status, and performance characteristics of physical objects [1]. Through Digital Twin technology, the actual driving school vehicles can be combined with their virtual counterparts in the digital realm, enabling real-time monitoring, analysis, and optimization [2]. This technology not only provides an accurate understanding of the driving school vehicle’s status but also simulates various driving scenarios, aiding students in practicing driving within a safe environment and effectively enhancing their driving skills.

In China, satellite navigation systems such as GPS and Bei Dou have been widely applied in the field of vehicle positioning for tasks such as vehicle navigation, real-time positioning, and path planning. With the expansion of vehicle-to-everything (V2X) technology, the integration of intelligent vehicle positioning systems and the Internet has become increasingly seamless in China. Vehicles can transmit real-time location information through the network, providing more real-time data for drivers and systems. In foreign countries, high-precision positioning technologies like Real-Time Kinematic GPS (RTK-GPS) are extensively utilized, offering sub-meter or even higher accuracy positioning services, particularly applicable to areas like autonomous driving [3]. The application of intelligent vehicle positioning systems in foreign countries has grown more prevalent in fields such as autonomous driving and vehicle-to-vehicle communication, enabling vehicle-to-vehicle communication and location information to facilitate fleet coordination and traffic flow optimization [4].

In reference [5], a collaborative underwater network for an ocean observation system was deployed using Digital Twin methodology. Reference [6] explores the necessity of Digital Twins as decision-making tools from a demand perspective. In conclusion, Digital Twin technology holds substantial practical significance for advancing the development and evaluation of intelligent vehicle functions. However, there are relatively few research reports on the specific implementation, validation, and applications of Digital Twin technology in intelligent vehicles. This paper aims to investigate an intelligent driving school vehicle system based on Digital Twin technology. By integrating physical vehicles with Digital Twin models, real-time vehicle monitoring, vehicle positioning, and driving training optimization can be achieved. This study will delve into three aspects: multi-source data acquisition, local network establishment, and data visualization interface design. Application testing will be conducted on actual driving school vehicles, aiming to provide innovative solutions for driving school vehicle management and training. This research is expected to contribute new ideas and approaches to the development and application of intelligent driving school vehicle systems.

2 Multi-Source Data Acquisition for Intelligent Driving School Vehicles

The vehicle sensor data acquisition module utilizes both serial ports and Controller Area Network (CAN) to collect and parse the serial output information from the P3DU receiver unit in the driving school vehicle’s electronic control system. This includes location data and velocity data. Additionally, it collects and interprets CAN messages from the On-Board Diagnostics (OBD) interface, gear position sensor (not equipped in automatic transmission training vehicles), and brake/clutch/accelerator pedal angle sensors (not equipped in automatic transmission training vehicles), thereby obtaining signals from various sensors.

The CAN messages from the driving school vehicle can be categorized into three types: OBD interface information, CAN messages from the gear position sensor, and CAN messages from the brake/clutch/accelerator pedal angle sensors. The CAN messages from the OBD interface encompass status information such as handbrake, turn signals, and driver’s door. For the RTK-GPS system, data collection involves obtaining the $GPHCD position information from the P3 receiver unit’s serial port 1, as well as the $GNVTG data from serial port 2.

3 IoT-Based Data Transmission Architecture

This section aims to propose an IoT-based data transmission architecture to meet the functional requirements of the smart driving school coach robot hardware system. By establishing a local area network (LAN) communication between the Data Monitoring Center and the smart driving school coach, real-time monitoring of vehicle status information, training data storage, and audio–video communication functionalities are achieved. Specifically, the network architecture employs the AP + AC mode, with the Data Monitoring Center serving as the core to establish a comprehensive LAN for centralized management and control of the smart driving school coach.

To realize the data transmission architecture, this paper adopts the AP + AC networking mode to build the driving school's local area network. As shown in Fig. 1, the router is directly connected to the Internet, and the switch is used to assign IP addresses to various connected devices in the lower layer. The AC functions as a centralized management device, overseeing the configuration of parameters such as IP addresses, channels, and power for the APs. Terminal devices access the Internet through the wireless network generated by the APs.

Fig. 1
An illustrated flow diagram of a technology system in a coach car. It depicts the connection between various devices like the car roof computer, PAD, and face recognition through access points and switches.

AP + AC networking mode

Full size image

To establish an internal local area network (LAN) for the intelligent driving school, this study employs router devices that support Access Controller (AC) functionality, connecting multiple Access Points (APs) via switches to create a wireless network within the training area. The instructor vehicles connect to the wireless network within the training area, enabling local area network communication between the vehicles and the data monitoring centre.

4 Designing a Vehicle Data Visualization Interface

To achieve real-time monitoring and positioning of vehicle status within the driving school training area, this section presents an interface design solution based on the ThingJS platform. The interface functionality encompasses displaying vehicle operational status, visually presenting vehicle travel data, and providing real-time vehicle location tracking. Through this interface design, driving school administrators can promptly grasp vehicle operational status and location information, offering timely and effective data support to instructors.

4.1 Vehicle State Presentation

The in-vehicle pad student training interface establishes a WebSocket connection with the onboard computer to receive vehicle operating status information, including driving speed, steering wheel angle, left and right turn signal status, clutch pedal angle, brake pedal angle, throttle pedal angle, driver’s door status, seatbelt status, handbrake status, and ignition status.

4.2 Vehicle Location Display

The objective of the GPS coordinate positioning module is to take the positioning information received by the vehicle sensor data acquisition module, calculate it in the coordinate system of the training area, and determine the position coordinates of vehicle key points (such as headlights and wheels) based on the vehicle's heading angle. This is done to determine the current training subject for the student, generate corresponding instructions for the start or completion of the training subject, and transmit them to the onboard pad. Figure 2 shows an example of the in-vehicle Pad trainee training interface.

Fig. 2
A screenshot of a driving simulation interface. It highlights various controls and information for learning purposes. The interface has various controls and provides real-time feedback, enhancing the understanding of driving dynamics in a safe, simulated environment.

Vehicle and training area Three-Dimensional models

Full size image

5 Application Examples of Intelligent Driving Instructor Car Systems

Taking the Intelligent Driving Instructor Robot Project at a Driving School as a Case Study, Demonstrating the Application Achievements of Intelligent Vehicle Systems in the Context of the Intelligent Driving Instructor Robot Project.

5.1 System Deployment

By setting up a local area network, communication between the instructor's vehicle and the data monitoring center is established, allowing the transmission of instructor vehicle status information to the centre. Sensors. The OBD-CAN interface, gear position sensor, and brake/clutch/throttle pedal angle sensors are installed at corresponding positions within the driver's cabin to achieve the collection of vehicle status information. Auxiliary brake control system. The auxiliary brake control system is placed on the vehicle roof. It comprehensively analyzes ultrasonic radar signals, millimeter-wave radar signals, speed information, reverse light signals, ignition signals, handbrake signals, shielded brake signals, etc. It then executes corresponding control strategies to ensure the safety of the driver. Figure 3 illustrates the main hardware devices added to the vehicle hardware modification.

Fig. 3
An illustration of a car highlights the hardware modification including I R M, H M I, ultrasonic probe, wireless network antenna, dual-antenna R T K G P S, onboard controller, seat belt, alarm switch, door limit switch, and gear position sensor.

Driving school vehicle hardware modification

Full size image

The main design section showcases the vehicle's positional status, depicting its location within the driving school training ground model. Additionally, auxiliary practice buttons and information display boxes are integrated into the surroundings. Through the frontend interface's vehicle positioning display, vehicle status presentation, and auxiliary function design, learners can observe the vehicle's operational location and status during the driving skill learning process. This provides a more intuitive understanding of the impact of driving manoeuvres on vehicle control. Figure 4 demonstrates a test screen after the hardware and software systems have been deployed.

Fig. 4
Two photos of a driving training interface. One is an in-car digital screen providing driving data. The other is an exterior view of a person holding a tablet that mirrors the in-car interface, with a car and driving course in the background.

In-Car and exterior pad training interfaces

Full size image

5.2 Data Analysis and Processing

Through the intelligent driving school car system, vehicle positioning data is collected and subjected to coordinate system transformation processing. As a result, coordinates in the training ground's coordinate system are obtained. As shown in the figure, the initial data obtained is relatively coarse, requiring further smoothing and approximation for display on the front-end interface. Due to the high real-time requirements of vehicle positioning, three data processing methods are proposed.

$$x = 0.25 \times x_{{{\text{t}} - 2}} + 0.25 \times x_{t - 1} + 0.5 \times x_{t}$$
(1)
$$x = round(x_{t} \times 50)/50$$
(2)
$$x = round(x_{t} \times 12.5)/12.5$$
(3)

Here, x represents the coordinate value processed at the current time step, \(x_{t}\) denotes the raw data of the coordinate value at the current time step, \(x_{t - 1}\) corresponds to the raw data of the coordinate value at the previous time step, \(x_{t - 2}\) represents the raw data of the coordinate value received two time steps ago, and ‘round’ indicates rounding to the nearest integer. The subsequent method 1, method 2 and method 3 correspond to data processing approaches in Formulas (1), (2) and (3), respectively. Dividing vehicle status into two categories, motion and stationary, based on velocity information.

For the motion scenario, taking the example of original data during a vehicle's movement on a straight road segment, the x-axis data gradually increases while the y-axis data is subjected to three different processing methods successively to create plots. The resulting graph is shown in Fig. 5 below. From the graph, it is evident that the curve obtained from processing method 3 fits the original data less accurately compared to processing methods 1 and 2. The smoothness of the curve obtained from processing method 2 is not as good as that of processing method 1. Therefore, processing method 1 is more suitable for data processing under motion status.

Fig. 5
A line graph plots y in meters versus x in meters. Four fluctuating lines labeled methods 1, 2, and 3, and raw data start between 14.7 and 14.75 on the y-axis, increase slightly, decrease to the minimum, and increase slightly at the end.

Coordinate data line graph during vehicle motion

Full size image

In the case of vehicle at rest, taking the original data of the vehicle stationary at a certain point in the ignition state as an example, data processing was conducted using three different methods for both x and y-axis data. The resulting scatter plots are shown in the figure below. It can be observed from Fig. 6 that method 1 still yields a set of points with significant data fluctuations, method 2 ultimately yields 4 coordinate points, and method 3 yields 2 coordinate points. Therefore, for the stationary state, data processing method 2 is more suitable.

Fig. 6
A scatterplot of stationary coordinate data plots y versus x. The plots for raw data, method 1, method 2, and method 3 linearly rise. The plots are dense between 10.542 and 10.555 on the x-axis and 38.52 and 38.54 on the y-axis.

Coordinate data scatter plot during vehicle standstill

Full size image

6 Conclusions

  • This paper introduced the hardware and software components of an intelligent vehicle system, realizing the collection of multi-source data from sensors, its transmission through the network to the computer, and subsequent interface display;

  • Combining with the ThingJS platform, this study developed a frontend interface, focusing on vehicle positioning, vehicle status display, and adaptive functionality for drivers, facilitating the learning of driving skills by learners;

  • Through practical hardware modification and system testing on actual driving school vehicles, this paper processed the received coordinate data for display on the frontend interface. After testing, suitable data processing methods were selected for vehicles in motion and stationary states, respectively.