Abstraction and Persistence: Macro-Level Guarantees of Collective Bio-Inspired Teams under Human Supervision

Abstract

It can be difficult for humans to control large numbers (100-200) of robots performing coordinated tasks. Organizational constraints may be imposed that allow a human to issue commands or plays that dictate collective behavior, but naive hierarchical approaches can suffer from robustness issues in cases where key robots or communication channels are compromised. Bio-inspired models can create robust and purposeful collective behavior that is implemented in a completely decentralized organization. Unfortunately, although robust, it may be difficult for a human to influence collectives in a way that the collective is responsive to the human input. We present results from an iterative modeling exercise that systematically relates how individual agents can be influenced by humans to ensure macro-level behaviors of the collective. I.