Dr Kai Fung Chu

Strategic Themes 

Predictive Traffic Data Analysis 

Forecasting traffic data according to its high temporal-spatial resolution and real-time characteristic using deep learning models. 

Multi-agent Traffic Management 

Management of large-scale vehicle fleets and road infrastructure for pre-defined objectives using reinforcement learning. 

Trustworthy Artificial Intelligence for Intelligent Transportation Systems 

Adversarial machine learning against cyber-physical attacks to applications of intelligent transportation systems. 

Research Project 

Title: Control and implications of mixed autonomous vehicle-infrastructure in a heterogeneous multi-agent system framework 

Theme: Automation & robotics 

Abstract:

Connected autonomous vehicles (CAVs) will be prevalent given the technology maturity and government promotion. Transportation system administrators and constructors should be prepared to leverage the controllability and potential of CAVs when they gradually permeate the roads in the near future. Current studies for CAVs usually consider either the control of a single CAV with human-driven vehicles (HVs), or collectively control among CAVs only. Moreover, their interactions with intelligent road infrastructure are investigated separately, which oversimplifies the heterogeneity among CAVs, HVs, and intelligent road infrastructure in the near future transportation system. Numerous challenges of heterogeneous multi-agent transportation systems, such as their interactions, partial controllability, and implications, are not addressed. This project will study the control and implications of a heterogeneous multi-agent transportation system mixed with CAVs, HVs, dynamic reversible lanes, and intelligent traffic lights. First, with awareness of the high spatial-temporal resolution and real-time characteristics of transportation systems, an efficient heterogeneous data fusion and multi-agent modelling framework will be developed. Second, optimal control policies for the heterogeneous multi-agent transportation system satisfying the safety requirements will be developed. Third, the opportunities and barriers to practical implementation and the implications for society and governance will be analysed. Through this project, we will push forward the technologies and insights of CAVs at dynamic reversible lanes and intelligent intersections and move towards a safer and more efficient system whereby CAVs can be highly leveraged. 

Alignment with SDG: Goal 11 (Sustainable cities and communities) 

Project TRL: TRL 3 (Experimental proof of concept) 

Industry secondment needs: Real-world or simulated experimental platform for multiple autonomous vehicles 

e_sdg_logo_without_un_emblem_square_web.png.png

e-web-goal-11.png