A Feasibility Study of a Traffic Supervision System Based on 5G Communication

Area of Science:

• Robotics and Artificial Intelligence
• Transportation Engineering
• Networked Systems

Background:

• Current autonomous vehicles operate in an ego-vehicle manner, relying solely on onboard sensors for navigation.
• Mobile network connections are typically limited to non-critical functions, not core driving tasks.

Purpose of the Study:

• To propose and evaluate a connected autonomous driving system utilizing a central road supervisor.
• To demonstrate the benefits of centralized coordination for enhanced decision-making in autonomous vehicles.

Main Methods:

• Vehicles transmit real-time data (position, velocity) to a road supervisor.
• The road supervisor processes aggregated data to generate coordinated, recommended trajectories for all supervised vehicles.
• System performance validated using a popular traffic simulator and ultra-low-latency 5G networks.

Main Results:

• The connected system demonstrated efficient and safe operation in simulations.
• Centralized planning by the supervisor enabled more sophisticated decisions than individual vehicle planning.
• Feasibility of offloading trajectory planning to the network via 5G was confirmed.

Conclusions:

• A connected autonomous driving system with a road supervisor offers significant advantages over traditional ego-vehicle approaches.
• Network-based trajectory planning, especially with 5G, is a viable strategy for future autonomous driving systems.