A graduate course on the data, models, and control systems behind modern mobility — from sensing and prediction to adaptive signal control and the governance of safety-critical AI on real road networks.
Curriculum
Overview
This module examines how cities sense, predict, and manage the movement of people and vehicles, and what changes when machine learning enters a safety-critical control loop. Students leave able to read the research literature critically and to reason about the gap between a working prototype and a deployed system thousands of people depend on daily.
Learning outcomes
By the end of the module, students will be able to:
- Explain the main data sources and their failure modes in real transport networks.
- Build and evaluate short-term traffic prediction models, including spatiotemporal deep-learning approaches.
- Compare rule-based and learned control strategies, and design appropriate fallback behaviour.
- Critically assess the governance, safety, and ethical questions raised by deploying AI in public infrastructure.
Assessment
| Component | Weight |
|---|---|
| Practical assignment (prediction pipeline) | 30% |
| Group project & presentation | 30% |
| Final examination | 40% |
Prerequisites
A working knowledge of Python and introductory machine learning (or equivalent). The first lecture revisits the essentials.
Lecture slides
- 01 Course overview & the mobility data landscape Download PDF ↓
- 02 Sensing the network: loops, cameras, GPS, probes Download PDF ↓
- 03 Traffic flow theory & fundamentals Slides coming soon
- 04 Short-term forecasting: classical methods Slides coming soon
- 05 Spatiotemporal deep learning for prediction Slides coming soon
- 06 Adaptive signal control I: rule-based systems Slides coming soon
- 07 Adaptive signal control II: learned controllers Slides coming soon
- 08 Robustness, fallback, and graceful degradation Slides coming soon
- 09 Governance & ethics of safety-critical AI Slides coming soon
- 10 Case studies & guest lecture Slides coming soon