Interactive Online Choreography for a Multi-Quadrotor System

Abstract

We present a full system for online control of multiple quadrotors. Rather than following scripted directions, our system interprets user instructions and modifies quadrotor behavior online, allowing a user to issue instructions at any time to easily control and modify flight. Handling multiple quadrotors is done through grouping to allow a user to reduce a number of individual robots into scalably sized collectives as needed. Instructions are specified from a simple enumeration of potential action combinations governing formation shape, flight speed and style, and overall group tasking (travel to target, aim spotlight, etc), which combinatorially yield over 8000 possible behaviors. The system interprets all user commands into dynamically feasible, non-colliding trajectories, and results are shown for over 100 behavior instructions issued online to a multi-quadrotor hardware system used for a theatrical performance.

 

Authors

  • Ellen A. Cappo, PhD Student, Carnegie Mellon Robotics Institute

  • Arjav Desai, PhD Student, Carnegie Mellon Robotics Institute

  • Nima Dehghani, Masters Student, Carnegie Mellon School of Art

  • Ali Momeni, Associate Professor, Carnegie Mellon School of Art

  • Nathan Michael, Associate Research Professor, Carnegie Mellon Robotics Institute

 
 

Citation

“Interactive online choreography for a multi-quadrotor system,” in Workshop on Human-Robot Interaction for Small and Personal Unmanned Aerial Vehicles, Robotics Science and Systems, Ann Arbor, MI, USA, 2016.