V2X Network-Enabled Traffic Analysis and Smart Traffic Signal Control for Large Traffic Networks

An efficient people-mover system is the backbone of a smart city. It is vital to develop an integrated and sustainable transport system that meets the diverse needs of a burgeoning population.

Despite the advancement of transport related technologies, the main challenge, of ensuring safety, comfort, affordability in terms of time and money, environmental footprint, and social impact, remains.

This project addresses two key problems from a systems and control perspective – discovering and understanding people’s travel needs and commute patterns at a societal level via traffic analysis and prediction, and using traffic signal control, which is essentially a group control mechanism, to enhance safety, comfort and affordability of daily travels in a large complex traffic network. 

Systems and control ideas have long been used in developing solutions to address safety, comfort, and affordability in mobility. First principle models are getting more difficult to obtain, owing to the sheer scale and complexity of a mobility system. New information technologies are bringing people directly into the decision loop that requires new design procedures. Integrating people such as pedestrians and passengers, system infrastructure and data to form a real-time smart mobility system that can significantly enhance safety and travel comfort, while appropriately reducing costs, has been the new focus of research and development.

By developing machine learning techniques to predict spatial and temporal evolution of traffic congestions and utilising C-V2X technologies, the drawbacks of existing traffic signal control systems can be addressed. To tackle uncertainties in traffic flow, such as sporadic disturbance caused by traffic accidents and other abnormalities, at a scale that involves thousands of junctions and tens of thousands of links, this project aims to develop a real-time signal control system. To reduce computational complexity, a structural solution will be developed where a large network is partitioned into several small sub-regions, and each sub-region is equipped with a local traffic controller that communicates with other regional controllers to achieve a good global control solution. Figure 1 depicts such a distributed signal control architecture and its interaction with vehicle control and bus management.

Real-time traffic signal control will play a major role in several areas of traffic management, such as regular or on-demand bus management during peak hours, management of platoons of vehicles for passengers or goods, and traffic control for emergency situations.

Using the V2X infrastructure on the NTU campus to develop a testbed to illustrate developed methods and tools, this project targets to create a toolbox for traffic modelling and parameter estimation, congestion forecast, condition prediction and travel guidance, and signal scheduling, to enable the management of public buses and platoons of vehicles, as well as emergency handling.