The studies below are currently under review at the conference or are works in progress. Therefore, only the general direction is described, excluding detailed titles.
Does the current road congestion index accurately reflect the impact on the network? If not, how can we fix it?
Submitted to 103rd TRB, working on journal submission
Do we really need heavy models to predict traffic phenomena? How can we achieve high accuracy while increasing the interpretability of the model?
Paper writing is in progress
How can we leverage the data from various sources? How can we interpret the function of each data?
Submitted to 103rd TRB, working on journal submission
Is the current reinforcement learning framework sufficient to imitate the real-world behavior of drivers? If not, how can we formulate the problem?
Submitted to 103rd TRB and journal
Current traffic simulations take a long time to reach equilibrium. Is there really no way to accelerate this?
Submitted to 103rd TRB