Using Big Data and A.I. for urban infrastructure risk assessment

Almost all future population growth will take place in urban areas, particularly in the Global South. These cities face common challenges of: 1) informality; 2) rapid change outstripping data collection, and; 3) limited capacity to manage risk. Resultantly, it is often difficult to apply techniques of risk analysis from the Global North. Additionally, many Global South cities are in hazard-prone regions.

This PhD will explore how to translate big data and A.I approaches to assess urban infrastructure risk in the Global South. The work will build upon previous research developing a risk map of infrastructure vulnerabilities to natural hazards based off classification of urban areas into 17 infrastructure typologies (‘urban textures’) using a combination of remote sensing, fieldwork, and expert judgement of Google Streetview. This has been used to create maps of Nairobi, Karonga, Niamey showing how different areas might be affected differently by single and interacting hazards. 

Leveraging AI, this project aims to automatically extract infrastructure typologies from Streetview images, and big datasets of remote sensing imagery to zone cities in order to map all cities of the Global South into levels of infrastructure risk. Depending on your interests, topics include: a) how ‘fit for purpose’ AI is for slums; b) what AI platforms are most effective; c) an automated workflow for assessing temporal change, or; d) changes associated with climatic and/or socio-environmental change.