Efficient Evaluation of COVID-19 Safety for Interior Layouts

Andrew Bolton


Supervised by Bailin Deng; Moderated by Jing Wu

Due to COVID-19, there is a need to design new indoor workspaces or modify existing ones in a way that is compliant with COVID safety guidelines. For example, the occupants should be able to easily move to different parts of the space while maintaining social distancing. For this purpose, it is important to have a tool that can quickly evaluate the performance of an interior layout based on the ease of movement within the space subject to social distancing constraints. This project will develop such a tool using agent-based simulation and neural networks. First, agent-based simulation will be used to simulate occupants who are social-distancing-aware, and to determine how easy it is for an occupant to move around within a given interior layout. A large number of simulations will be run using different layouts to collect training and testing data for a neural network. Using these data, we will train a network to predict the performance of an arbitrary interior layout.

Initial Plan (08/02/2021) [Zip Archive]

Final Report (28/05/2021) [Zip Archive]

Publication Form