Grounding Language in Multi-Perspective Referential Communication

Zineng Tang, Lingjun Mao, Alane Suhr

We introduce a task and dataset for referring expression generation and comprehension in multi-agent embodied environments. In this task, two agents in a shared scene must take int…