Can LLM agents reproduce human travel choices?
I build behaviorally aligned persona agents that extend small paired RP–SP surveys into population-scale experiments for autonomous shuttles and DRT.
Travel behaviorMobility AI · Multi-Agent Systems · Travel Behavior
AI for understanding
human mobility.
I’m Chanwoo Moon, an M.S. student at KAIST’s TUPA Lab, working at the intersection of transportation engineering, behavioral modeling, and foundation models.
People
MOBILITY
INTELLIGENCE
INTELLIGENCE
Systems
From human behavior
to safe mobility decisions.
Academic affiliations
Research directions
Three connected research questions.
How can vehicles and infrastructure reason together?
I study cooperative world models for autonomous driving at signalized intersections, linking physical context with decision-making.
Physical AIHow do we keep mobility AI grounded?
I combine behavioral constraints, retrieval, held-out human responses, and reproducible evaluation to test where AI systems succeed—and fail.
Trustworthy AICurrent work
LLM agents for new-mobility demand.
Research thread
From paired surveys to city-scale social-response experiments
My current work investigates whether persona-based LLM agents can serve as synthetic survey respondents, represent diverse traveler profiles, and support demand analysis when observations for new services are scarce.
Read the research overviewBehavior
Agents
Mobility
Evaluation
Now
Current milestones
Recent steps across mobility research, autonomous driving, and applied AI.
-
Cooperative world-model research launchedStarted a Physical AI research thread for vehicle–infrastructure cooperation at signalized intersections.
-
Joined KAIST TUPA LabBegan the M.S. in Mobility at the Cho Chun Shik Graduate School of Mobility.
-
Graduated from Yonsei UniversityCompleted Urban Planning & Engineering with a minor in Applied Statistics and received major capstone and transportation honors.