Abstract

Multiagent Hierarchical Learning from Demonstration
Multiagent Hierarchical Learning from Demonstration
Keith Sullivan
I introduce a learning from demonstration system, called Hierarchical Training of Agent Behavior (HITAB). In HITAB, agents learn a hierarchical finite state automata (HFA) represented as a Moore machine where individual states correspond to agent behaviors or another HFA. HITAB allows rapid training of both single agent and multiagent behaviors.