AUGMENTED URBAN SPACE MODELING LABORATORY

Embodied Spatial Intelligence

AI THAT UNDERSTANDS, PREDICTS AND ACTS IN CITIES, ON EARTH AND IN SPACE

Our researchJoin the lab

RESEARCH

Structured Spatial World Models

Our world models are structured. They represent the world as 3D geometry, objects, relations and physics, not just pixels, so that robots, drones and satellites can reason about the world and act in it. What they learn from acting flows back to update the model.

  • Urban World ModelsGraph models that understand and predict robots and mobility in cities
  • Scan-to-ModelInfrastructure models with engineering knowledge, built from scans
  • Embodied AutonomyRobots, drones and cars that sense, map and act in real time
  • Physics-Informed Earth AIAI guided by physics that emulates Earth systems
  • Space Situational AwarenessSpace in the loop: detecting and tracking objects in orbit

Our research

Research in Motion

Urban World Models Digital Twins: Simulating Robot–Human Sidewalk Interactions
Scan-to-Model Power Line Network Modeling and Single Tree Detection from Airborne Point Clouds
Embodied Autonomy Track Extraction from New York Subway Video (TPE-Net)

All videos

News & Events

Defence · Sep 21, 2026 Dr. Mohammadjavad Ghorbanalivakili successfully defends his Ph.D. dissertation Defence · Aug 20, 2026 Dr. Afnan Ahmad successfully defends his Ph.D. dissertation Event · Sep 24, 2026 Prof. Gunho Sohn Joins the Panel at Industry Night: Computing & Mechanics 2026 People · Sep 2026 Welcome to our new members

All news

Publications

YUPU YUPU ECCV Workshops, 2026 GeoPriorPC GeoPriorPC CVPR Workshops, 2026 Reasoning-guided Ego-path Segmentation Reasoning-guided Ego-path Segmentation ISPRS Archives, 2026 Plume Rise from Images with Deep Learning Plume Rise from Images with Deep Learning Geoscientific Model Development, 2026

Read More

AUSM Lab members, 2026
AUSM Lab members, 2026