Policy, Infrastructure, and Trustworthy Mobility AI

2021 - Present | Decision Support for Electrification, Resilience, and Safe AI Mobility Systems

Project Overview

This direction translates mobility AI into operational policy decisions. It combines market-level EV policy modeling, post-disaster logistics optimization, regulation-aware autonomous decision systems, and trustworthy multimodal mobility intelligence.

Motivation

Transportation decisions require more than predictive accuracy. They need interpretable models that can be audited, stress-tested, and mapped to concrete policy levers, infrastructure plans, and safety constraints.

Goal

Deliver decision-grade mobility intelligence that supports electrification strategy, disaster-response planning, and trustworthy AI deployment in high-stakes transportation settings.

Direction Narrative

Policy sensitivity for EV transition

EV adoption and charging deployment are modeled as a two-sided market feedback system, enabling quantitative policy scenario testing under infrastructure constraints.

Resilience operations for post-disaster recovery

Queue-aware scheduling models are used to optimize debris removal under uncertain and dynamically evolving disaster conditions.

Trustworthy AI for safety-critical mobility systems

Regulation-aware reasoning and multimodal benchmark datasets support interpretable autonomous decision pipelines and more robust cooperative mobility intelligence.

Semantic and uncertainty-aware mobility inference pipeline — Example inference pipeline emphasizing uncertainty handling and semantic context integration.

Results

EV policy modeling in LA County estimates that a 1% increase in EV charging stations is associated with around 0.35% increase in EV sales.
Debris logistics analysis indicates potential reduction from around 742 days to around 263 days under 24-hour operation scenarios.
DriveReg contributes an interpretable regulation-aware decision framework for autonomous driving compliance and safety reasoning.
V2X-Real contributes a large-scale cooperative perception benchmark for real-world multimodal V2X research.

Impact

This direction strengthens evidence-based transportation policy and operations by making AI outputs explainable, scenario-aware, and directly tied to implementation decisions.