Crowdsourced Multirobot Coordination


We investigate possible approaches to modelling human group behavior and coordination when solving collective tasks. We conduct experiments on groups of humans and study the group behaviors that emerge. The study of such behaviors allows us to mathematically express models that present the human decision-making process which can then be used as algorithms for distributed coordination of large groups of robots to solve complex distributed problems.

The video below demonstrates the animated data collected from one of the seven tasks conducted in our recent investigation. Here, the participants were asked to form a rectangle of a unified color using the local and global information available to them through their GUIs.

We investigate distributed problem-solving in human groups over a diverse range of collective tasks that have applications to robotics, including: multirobot target tracking, formation, objects clustering, and distributed optimization. We aim to devise a generic protocol for learning microscopic models of human group behavior in complex distributed collective tasks.


  • Arash Tavakoli
  • Haig Nalbandian
  • Nora Ayanian


This project is supported by NSF CAREER IIS-1553726.

Related Publications

Refereed Conference Papers and Abstracts

  • A. Tavakoli, H. Nalbandian, and N. Ayanian. "Crowdsourced Coordination Through Online Games", in ACM/IEEE Intl Conf. on Human Robot Interaction Late Breaking Reports, 2016.
    [ PDF Preprint, BibTeX ]

Workshop Papers

  • A. Tavakoli and N. Ayanian. "Multirobot Coordination by Multiplayer Games", in International Conference on Robotics and Automation (ICRA) Fielded Multi-Robot Systems Operating On Land, Sea, and Air Workshop, 2016.

Preprints and Papers Under Review

  • A. Tavakoli, H. Nalbandian, and N. Ayanian. "Multiplayer Games for Learning Multirobot Coordination Algorithms". arXiv preprint arXiv:1604.05942. 2016.
    [ PDF Preprint, arXiv, BibTeX ]