The paper "VoicePilot: Harnessing LLMs as Speech Interfaces for Physically Assistive Robots" by Akhil Padmanabha, Jessie Yuan, Zulekha Karachiwalla, Carmel Majidi, Janavi Gupta, Henny Admoni, and Zackory Erickson from Carnegie Mellon University presents a framework for integrating Large Language Models (LLMs) as speech interfaces for assistive robots. The authors focus on physically assistive robots, particularly a feeding robot named Obi, to enhance the well-being and independence of individuals with motor impairments or other disabilities. The framework is developed iteratively through three stages of testing, culminating in an evaluation with 11 older adults at an independent living facility. The first stage involves developing an initial framework based on existing literature, followed by pilot testing with lab members. The second stage refines the framework through user feedback and demonstrates it at a disability awareness event. The final stage evaluates the system with older adults, leading to the development of design guidelines. Key components of the framework include Environment Description, Robot Functions, Function Applications, Code Specifications, Safety, Robot Variables, Instructional Materials, User Control Functions, and Feedback. The authors also present design guidelines based on the study, emphasizing customization, multi-step instruction, consistency, comparable time to a caregiver, and social capability. The study shows that the LLM-based speech interface for the Obi robot was well-received by participants, with most completing tasks within three attempts and finding the interface easy to learn and control. The authors conclude that their framework and guidelines can enhance the user experience and provide valuable insights for researchers, engineers, and product designers working on assistive robots.The paper "VoicePilot: Harnessing LLMs as Speech Interfaces for Physically Assistive Robots" by Akhil Padmanabha, Jessie Yuan, Zulekha Karachiwalla, Carmel Majidi, Janavi Gupta, Henny Admoni, and Zackory Erickson from Carnegie Mellon University presents a framework for integrating Large Language Models (LLMs) as speech interfaces for assistive robots. The authors focus on physically assistive robots, particularly a feeding robot named Obi, to enhance the well-being and independence of individuals with motor impairments or other disabilities. The framework is developed iteratively through three stages of testing, culminating in an evaluation with 11 older adults at an independent living facility. The first stage involves developing an initial framework based on existing literature, followed by pilot testing with lab members. The second stage refines the framework through user feedback and demonstrates it at a disability awareness event. The final stage evaluates the system with older adults, leading to the development of design guidelines. Key components of the framework include Environment Description, Robot Functions, Function Applications, Code Specifications, Safety, Robot Variables, Instructional Materials, User Control Functions, and Feedback. The authors also present design guidelines based on the study, emphasizing customization, multi-step instruction, consistency, comparable time to a caregiver, and social capability. The study shows that the LLM-based speech interface for the Obi robot was well-received by participants, with most completing tasks within three attempts and finding the interface easy to learn and control. The authors conclude that their framework and guidelines can enhance the user experience and provide valuable insights for researchers, engineers, and product designers working on assistive robots.