Technology
Coordination architectures for autonomous systems that degrade gracefully and scale without central control.
01 — AGENT SWARMS
Agent swarms
Software-agent collectives that decompose a mission into tasks, negotiate assignments, and re-plan as conditions change. Our research focuses on coordination protocols that remain stable as the number of agents grows.
Mixture-of-experts routing distributes specialized competence across the swarm, so no single agent has to model the whole problem.
02 — ROBOTIC SWARMS
Robotic swarms
Embodied multi-robot teams operating in contested or communication-limited environments. We study control and estimation methods that tolerate platform loss and intermittent links.
Validation runs from physics-based simulation to hardware-in-the-loop testing, keeping algorithm assumptions honest against real sensing and actuation.
The hardware is part of the research program: we are developing mass production of swarm platforms using state-of-the-art additive manufacturing, integrating structural power with electroactive actuation as the complementary novelty to our coordination research.
03 — COLLECTIVE INTELLIGENCE
Collective intelligence
Decision quality that emerges from local rules and structured information sharing rather than central computation. Research questions include consensus under noise, information value, and the limits of emergent behavior.
04 — DISTRIBUTED AUTONOMY
Distributed autonomy
Architectures with no single point of failure: authority, planning, and state estimation are distributed so the system degrades gracefully instead of failing abruptly.
We analyze failure modes quantitatively — availability, recovery time, and mission-assurance margins under platform attrition.
05 — MULTI-AGENT SYSTEMS
Multi-agent systems
The formal foundations: game-theoretic task allocation, distributed optimization, and consensus algorithms. We build on established multi-agent theory and validate departures from its assumptions in simulation.