Voice Command Recognition for Explicit Intent Elicitation in Collaborative Object Transportation Tasks: a ROS-based Implementation

Voice command recognition remains relatively unexplored in robotics, with limited insight into user acceptance and real-world performance. In this work we try to address this by offering multiple voice command recognition models encapsulated in a single publicly available ROS node ready to be used b...

Full description

Bibliographic Details
Authors: Domínguez Vidal, José Enrique, Sanfeliu, Alberto
Format: article
Status:Versión aceptada para publicación
Publication Date:2024
Country:España
Institution:Consejo Superior de Investigaciones Científicas (CSIC)
Repository:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/388298
Online Access:http://hdl.handle.net/10261/388298
https://api.elsevier.com/content/abstract/scopus_id/85188060982
Access Level:Open access
Keyword:Human-centered Studies
Human-Robot Interaction
Intention Understanding
Natural Language Processing
Voice Recognition
Description
Summary:Voice command recognition remains relatively unexplored in robotics, with limited insight into user acceptance and real-world performance. In this work we try to address this by offering multiple voice command recognition models encapsulated in a single publicly available ROS node ready to be used by the robotics practitioner. We tested its actual performance with 10 volunteers of different nationalities whose first spoken language is not English. The obtained accuracy in these tests varies between 93.14% and 95.63% depending on the number of considered commands and model size. Finally, we conducted a user study with 23 new volunteers performing a human-robot collaborative transport task to test whether humans are willing to use this type of system despite having a non-negligible delay and failure rate. In addition to improvements in parameters such as comfort and trust in the robot, 86.9% of the volunteers chose this system over a technically more robust one.