Projects

A new environment that merges both physical and social dynamics, forcing AI agents to use social skills to affect their physical world.

The paper proposes a computational method to score how tranquil, social, or explorative spaces feel by combining daylight, pathways, and visibility data. Tested on 20 apartments, it found people rate spaces uniformly while the algorithm detects variation, with stronger alignment among experienced participants.

In 2023, at Autodesk Research we partnered with The Bentway in Toronto to use wearable sensors and cameras to capture how people experienced the public space under Toronto's Gardiner Expressway. Participants joined "walkshops", recording video, audio, surveys, and environmental data, while algorithms analyzed what they saw and visualized anonymized facial contours. The project shows how participatory, data-driven methods can help design public …

AvatAR is an AR system for exploring human motion data in place, using 3D trajectories, posture-based avatars, and embedded visualizations of gaze and touch. It combines an AR headset and a tablet for both mid-air and touch control, making motion data interactive and immersive.

OccSim simulates occupant behaviors in 3D building models to assess virus transmission risks over time. Tested with multiple stakeholder groups, it helped identify specific hazards and informed decisions on safer building design, operation, and policy.

EagleView speeds collaboration analysis by pairing video with real-time proxemic visualizations and timeline queries, letting analysts quickly find and interpret key moments.