Abstract
In this paper, we created a remote mobile controlled and monitored vehicle system, which is used in the self-driving and self-parking cars for the last kilometer. The system consists of three main parts: the car-mounted terminal, which is installed in the car for receiving remote orders from the mobile application and transferring information of the car to the application; the windows server, which is run in one computer that is connected to the internet, the server works as an information exchange bridge, which means that it is connected to the mobile application and car-mounted terminal using TCP/IP protocol; the mobile application, which is run in an Android mobile phone. The purpose of the system is to realize completely self-driving. The key technologies of the system include CAN bus communication technology, which is widely used in the car for solving the problems of the increasing automobile electronic communication system; 3G DTU, which is integrated with TCP/IP protocol and supports WCDMA and GPRS mode for obtaining DNS server address and IP address automatically.
Introduction: Bridging the "Last Kilometer"
As autonomous vehicle technology matures, the "last kilometer"—the final transition where a driver exits the vehicle and lets it park itself—remains a significant practical and psychological hurdle. Users require a reliable "digital tether" to ensure the vehicle is performing its task correctly while they are away from the driver's seat.
Technically, establishing this tether presents a connectivity challenge. Both the vehicle (via cellular data) and the user’s smartphone operate on mobile networks where IP addresses change frequently. Establishing a direct, stable Peer-to-Peer (P2P) connection in this environment is notoriously difficult. Furthermore, the system must bridge the gap between high-level mobile applications and the low-level Controller Area Network (CAN) bus protocols used internally by vehicles.
System Architecture: The Client/Server Bridge
To overcome the limitations of dynamic IP addresses, the researchers proposed a three-tier Client/Server (C/S) architecture. This design utilizes a central Windows server with a static entry point to act as a stable information bridge.
The architecture comprises three primary components:
- The On-Board Device: Built around a Freescale MCU, this hardware acts as a node on the car’s CAN bus. It uses a 3G Data Transfer Unit (DTU) to communicate with the server, collecting vehicle data and translating incoming remote commands into CAN frames.
- The Windows Server (The "Bridge"): This is the heart of the system. Because it has a static IP, both the car and the mobile app can connect to it regardless of their own shifting network addresses. The server handles user authentication and routes data packets using the TCP/IP protocol.
- The Android Application: A custom-built app that integrates the Baidu Map API for trajectory tracking and a dashboard for real-time diagnostics.
Hardware and Communication Protocols
A vital aspect of the system is the translation between different communication layers. The on-board terminal must convert high-level TCP/IP packets from the 3G network into low-level CAN frames. To ensure data integrity, the team developed a specific communication protocol where data is packed into frames with defined headers, length fields, and footers.
Crucially, the software is designed to give "remote orders" the highest priority in the processing queue. This minimizes latency during critical maneuvers like emergency braking. On the hardware side, the authors highlight the necessity of using CAN transceivers with digital isolation circuits to suppress system noise, ensuring that high-frequency communication doesn't interfere with sensitive vehicle electronics.
Implementation and Results
The prototype successfully demonstrated end-to-end duplex communication, allowing for both data harvesting (monitoring) and command injection (controlling).
Key results included:
- Real-time Monitoring: The Android app successfully displayed dynamic parameters including velocity, battery State of Charge (SOC), voltage, and charging status.
- Geographical Tracking: Integration with Baidu Maps allowed for real-time coordinate updates, providing the user with visual confirmation of the vehicle's location during self-parking.
- Reliable Connectivity: The C/S architecture effectively bypassed the "roaming IP" problem, maintaining a stable link between the user and the vehicle.
Conclusion and Future Work
While the system provides a solid framework, it serves as a guideline for future development. Several areas remain for exploration, including quantitative latency testing for safety-critical operations and the implementation of robust cybersecurity protocols to prevent unauthorized hijacking of the vehicle's CAN bus. Additionally, future work must address network reliability in deep underground parking structures where cellular signals may be weak.
Summary updated