Autonomy and Interdependence in Human-Agent-Robot Teams

A prescriptive methodology to guide the design and analysis of human-agent-robot systems has been proposed. The domain for the experiment is Blocks World for Teams (BW4T), a simulation environment. The goal in BW4T is to stack colored blocks in a particular order. The task environment comprises nine rooms containing a random assortment of blocks, plus a goal area for dropping them off. Each player controls an avatar, which the player can move between rooms to pick up and drop off blocks. Each team had two players, a human and a software agent; humans control their own avatars and command their agent partners through an appropriate interface. To compare the effects of changing autonomy, different autonomy treatments were defined. The results indicate a very clear decrease in burden as autonomy increased. The robot idle time decreases as autonomy increases, but this doesn't necessarily equate to increased effectiveness.