Skip to main content Skip to navigation

Charlotte Roman

Charlotte graduated in 2022 with a PhD in Mathematics for Real-World Systems.


The focus of Charlotte's research is modelling decision-making in traffic in order to improve intelligent road systems.

My research questions are:

  • How does information distribution negatively effect expected travel times of travellers?
  • How would drivers make decisions about route choice when given traffic light cycles?
  • What is the optimal reward function for intelligent traffic lights?
  • Are intelligent traffic lights fair?
  • How does autonomous vehicle route control affect the Price of Anarchy?

My supervisors are Paolo TurriniLink opens in a new window and Colm ConnaughtonLink opens in a new window.

Publications

Research Interests

Algorithmic Game Theory, Congestion Games, Fairness, Bounded Rationality, Graph Theory, Multi-agent Reinforcement Learning, and AI Safety.

Education

  •  MSc in Mathematics of Systems (Distinction). Chosen optional modules include Algorithmic Game Theory, Mechanism Design and Alternative Games, and Computational Methods for Complexity Science. Involved in working on a group project, with a pedestrian simulation company Legion Ltd., on multi-scale modelling for simulation of crowds. My individual MSc project was on the topic of using superorganism decision making as a model of bounded rationality.
  • BSc in Mathematics at the University of Warwick (First class honours). This included modules in Game Theory, Combinatorics, Statistics, Graph Theory, Bifurcation Theory, Mathematical Economics, and Finance.

Warwick Seminar Tutor for:

Work Experience

Research intern at the Center for Human-compatible AI (CHAI) at the University of California Berkeley for the summer of 2020. Worked on a project on safety in sequential social dilemmas.

Graduate research intern at the pedestrian simulation software company Legion Ltd.

LinkedIn Profile: www.linkedin.com/in/charlotte-roman-8a3113a8/

Non-technical Research Summary:

Three Minute Thesis