Authors

Zhao Han and Daniel Hammer and Kevin Spevak and Mark Higger and Aaron Fanganello and Neil Dantam and Tom Williams

Venue

Frontiers in Robotics and AI - Human-Robot Interaction

Publication Year

2025
When collaborative robots teach human teammates new tasks, they must carefully determine the order to explain different parts of the task. In robotics, this problem is especially challenging, due to the situated and dynamic nature of robot task instruction. In this work, we consider how robots can leverage the Givenness Hierarchy to "think ahead" about the objects they must refer to so that they can sequence object references to form a coherent, easy-to-follow series of instructions. Our experimental results (n=82) show that robots using this GH-informed planner generate instructions that are more natural, fluent, understandable, and intelligent, less workload demanding, and that can be more efficiently completed.