An Adaptable, Safe, and Portable Robot-Assisted Feeding System

An Adaptable, Safe, and Portable Robot-Assisted Feeding System

March 11–14, 2024, Boulder, CO, USA | Ethan K. Gordon, Rajat Kumar Jenamani, Amal Nanavati, Ziang Liu, Daniel Stabile, Xilai Dai, Tapomayukh Bhattacharjee, Tyler Schrenk, Jonathan Ko, Haya Bolotski, Raida Karim, Atharva Kashyap, Bernie Hao Zhu, Taylor Kessler Faulkner, Siddhartha S. Srinivasa
The paper presents a robot-assisted feeding system designed to help individuals with mobility impairments feed themselves. The system emphasizes safety, portability, and user control, featuring comprehensive safety checks, the ability to be mounted on any powered wheelchair, and a custom web app that allows users to control the robot using their assistive devices. The system uses multi-modal online learning for bite acquisition and real-time mouth perception for bite transfer, ensuring adaptability and safety. It has been validated through multiple end-user studies and deployments, successfully feeding individuals with severe medical conditions. The hardware includes two robot arms (Kinova Gen2 and Gen3), a custom fork assembly, and a laptop with an Nvidia RTX 3060 GPU. The software is built on ROS2 and uses behavior trees and a React app for user interaction. Safety features include compliant hardware, control, anomaly detection, and emergency intervention mechanisms. The system is designed to be portable and user-controlled, allowing users to customize various aspects of the feeding process.The paper presents a robot-assisted feeding system designed to help individuals with mobility impairments feed themselves. The system emphasizes safety, portability, and user control, featuring comprehensive safety checks, the ability to be mounted on any powered wheelchair, and a custom web app that allows users to control the robot using their assistive devices. The system uses multi-modal online learning for bite acquisition and real-time mouth perception for bite transfer, ensuring adaptability and safety. It has been validated through multiple end-user studies and deployments, successfully feeding individuals with severe medical conditions. The hardware includes two robot arms (Kinova Gen2 and Gen3), a custom fork assembly, and a laptop with an Nvidia RTX 3060 GPU. The software is built on ROS2 and uses behavior trees and a React app for user interaction. Safety features include compliant hardware, control, anomaly detection, and emergency intervention mechanisms. The system is designed to be portable and user-controlled, allowing users to customize various aspects of the feeding process.
Reach us at info@study.space