Skip to content
Kooktae Lee edited this page Feb 22, 2026 · 2 revisions

D2OC: Density-Driven Optimal Control

Official Documentation & Knowledge Base

Welcome to the official repository for Density-Driven Optimal Control (D2OC). This project provides the implementation of the research published in IEEE Transactions on Systems, Man, and Cybernetics: Systems.

Unlike traditional coverage control, which focuses on reaching a static final robot distribution, D2OC introduces a dynamic, density-driven paradigm closer to finite-time ergodic control.


Why D2OC?

Most conventional coverage methods (e.g., CVT) are designed for stationary robot placement. D2OC redefines multi-agent coordination by treating the coverage problem as a dynamic process of matching time-averaged agent statistics to spatial density maps.

Key Research Contributions:

  • Dynamic Density Tracking: Moves beyond static placement to achieve continuous, ergodic-like coverage of nonuniform areas.
  • Optimal Transport Framework: Utilizes the L2 Wasserstein distance to mathematically guarantee that agent mass "flows" to represent the target density map over time.
  • Fully Decentralized Architecture: Computational cost remains scalable as agent numbers increase, with each agent using only local weight updates.
  • High-Fidelity Dynamics: Supports Quadrotor-inspired LTI dynamics (8-state models), ensuring the control logic is applicable to real-world robotic systems.

🛠 Getting Started

Experience the difference between static coverage and D2OC's dynamic tracking.

  • Interactive Demo: Watch how agents dynamically explore the density map.
  • MATLAB Implementation: The original research code for IEEE TSMC publication.
  • Python Implementation: Cross-platform version for modern robotics research.

How to Cite

@ARTICLE{seo2025density,
  author={Seo, Sungjun and Lee, Kooktae},
  journal={IEEE Transactions on Systems, Man, and Cybernetics: Systems}, 
  title={Density-Driven Optimal Control for Efficient and Collaborative Multiagent Nonuniform Coverage}, 
  year={2025},
  volume={55},
  number={12},
  pages={9340-9354},
  doi={10.1109/TSMC.2025.3622075}}