Journal of Assistive Robotics, 12(3), pp. 201--224, 2023

Abstract

Assistive robot arms are hard to control with low-dimensional interfaces such as joysticks. We model the user's goal as a hidden variable and let the robot assist in proportion to its confidence. A study with twelve participants shows faster task completion and lower workload than direct teleoperation.

BibTeX
@article{jones2023intent, author = "Jones, Judy and Stone, Sybil and Adams, Alice", title = "Inferring User Intent for Shared Autonomy under Uncertainty", journal = "Journal of Assistive Robotics", volume = "12", number = "3", pages = "201--224", year = "2023", doi = "10.5555/example.jar.2023.1203", abstract = "Assistive robot arms are hard to control with low-dimensional interfaces such as joysticks. We model the user's goal as a hidden variable and let the robot assist in proportion to its confidence. A study with twelve participants shows faster task completion and lower workload than direct teleoperation.", project = "sharedcontrol" }