RESEARCH
We connect AI, robots, autonomous vehicles, and people through collective intelligence.
Through game theory, control, and learning, we turn local decisions into coordinated outcomes—and collective goals into local action.
Intelligence is relational.
Intelligence matters through its relationships with others. We envision autonomous systems that understand the people and agents around them, navigate competing interests, and turn individual capabilities into shared progress. The goal is autonomy that contributes to the world it participates in.
Individual lane choices shape how efficiently traffic merges. This work models the strategic decisions of highway mainline vehicles near on-ramps, connecting self-interested behavior with aggregate traffic performance—a foundation for designing better interactions in mixed autonomy.
American Control Conference (ACC), 5376–5381 · 2020
Conference paper
A strategic AV changes more than its own trajectory: it changes how human drivers respond. This work models that feedback at weaving ramps and identifies performance plateaus and critical thresholds, showing when additional autonomy can—and cannot—improve traffic flow.
arXiv:2604.21941 · 2026
Preprint · Under review
MAPLE translates network-level congestion objectives into vehicle-level incentives through Pigouvian lane control. It connects the costs a lane decision imposes on others with decentralized control, aligning local actions with collective traffic efficiency.
8th Bridging Transportation Researchers Conference (BTR) · 2026
Conference contribution · Manuscript under review
Western Track Best Paper Award · 8th BTR Conference, 2026
Every route changes the network.
Mobility is a connected system: what happens in one place changes what becomes possible elsewhere. We envision networks that move people and resources efficiently, expand access to opportunity, and reconcile local needs with shared capacity. Our ambition is to make connectivity a source of collective benefit.
We model ride-hailing platforms, mixed fleets, customer patience, and congestion as one coupled equilibrium. The framework examines how routing freedom and fleet composition redistribute outcomes among companies, travelers, and the wider network.
arXiv:2512.10194 · 2025
Preprint · Under review
A unified toll-lane framework brings vehicle automation and passenger occupancy into the same policy decision. It studies who receives access, who pays, and how self-interested lane choices determine the outcome.
arXiv:2403.14011 · 2024
Preprint
When the world shifts, intelligence must move together.
When conditions change, collective capability should endure. We envision people, autonomous systems, and communities that can adapt together—even when information is incomplete and resources are strained. Our goal is to help societies respond to disruption while protecting the people who depend on them.
EVAC-AI brings people, AI agents, and transportation networks together for adaptive wildfire evacuation. We connect changing hazards with coordinated decisions, helping communities respond as conditions evolve—with vulnerable populations at the center.
NSF FIRE · $2M · 2026–2030
arXiv:2610.10946 · 2026
Preprint
arXiv:2608.22808 · 2026
Preprint
Traffic-network sensing is a resource-allocation problem: which locations deserve measurement when instrumentation is limited? This work examines the structure of optimal sensor placement, with submodularity as a lens for understanding the value of additional observations.
Transportation Research Part B: Methodological, 171, 29–43 · 2023
Journal article
Across scales. Toward shared progress.