On the Importance of Environments in Human-Robot Coordination
This addresses the need for better environmental design in human-robot interaction, though it is incremental as it builds on existing benchmarks like Overcooked.
The paper tackles the problem of understanding how environments affect human-robot coordination in collaborative tasks, proposing a framework for procedurally generating environments that are stylistically similar, solvable, and diverse, with results showing statistically significant differences in collaborative fluency metrics.
When studying robots collaborating with humans, much of the focus has been on robot policies that coordinate fluently with human teammates in collaborative tasks. However, less emphasis has been placed on the effect of the environment on coordination behaviors. To thoroughly explore environments that result in diverse behaviors, we propose a framework for procedural generation of environments that are (1) stylistically similar to human-authored environments, (2) guaranteed to be solvable by the human-robot team, and (3) diverse with respect to coordination measures. We analyze the procedurally generated environments in the Overcooked benchmark domain via simulation and an online user study. Results show that the environments result in qualitatively different emerging behaviors and statistically significant differences in collaborative fluency metrics, even when the robot runs the same planning algorithm.