Investigating subjective bias in machine-learning walkability systems

This project seeks to understand how pedestrians cultural background affect their perceptions of urban space, and how those perceptions are fed into modern urban planning systems. Walkability is a measurement of how easy it is for a pedestrian to move through space, and varies greatly throughout the world. Previous studies have shown walkability is highly subjective, and is influenced by ones cultural and economic background. One of the most cutting edge ways of measuring walkability utilizes machine learning to take data about a groups perception of different urban spaces and generalize those values to show larger patterns in walkability. However, most of theses studies do no take into account the aforementioned cultural subjectivity of the issue. This study seeks to gather data on peoples perceptions of these spaces from multiple cultural and economic backgrounds, and to use said data to create machine learning models that show how perceptions of walkabiliy vary between groups from different backgrounds. We are especially interested in gathering data from the pedestrians on the street in Pune, India and Montreal, Canada and comparing them.

Faculty Supervisor:

Raja Sengupta

Student:

Partner:

Symbiosis International University

Discipline:

Sociology

Sector:

Public Service, Policy, and Governance; Artificial Intelligence; Sustainability & the Environment

University:

McGill University

Program: