Stretch Community News - January 2024
Welcome to the Hello Robot monthly community update!
For the Hello Robot community, 2023 was a year filled with new publications, research projects, open-source code, and conferences around the world. Stretch even made the IEEE Spectrum Top Robotics Stories of 2023 list for research into assisting people with disabilities. Looking forward to the new year, we’re thrilled to keep discovering and sharing all the incredible work our users have been able to do with our robot.
This month, we have more exciting work done in real home environments - zero-shot autonomous pick-and-place in novel environments, and a head-worn robot teleoperation control interface - and Stretch assisting in projects from physical therapy to detecting plant diseases! Read on for details. If you’d like to see your work featured in a future newsletter, please let us know!
NYU and Meta released OK-Robot, an Open-Knowledge-based framework that offers an integrated solution for pick-and-drop operations without requiring any training! Combining a vision-language model with navigation and grasping primitives, OK-Robot can go from arriving in a completely novel environment to completing tasks in under ten minutes.
Robot-assisted physical therapy could eventually be a useful component for treating ailments like Parkinson’s disease. New research led by Georgia Tech, Emory School of Medicine, and TRI combines Stretch with a soft bubble end-effector, using this system as a physically and socially interactive robot for leading stretching exercise games with older adults with PD.
Researchers at Khalifa University used Stretch with a transformer-based model to achieve real-time diagnosis of tomato leaf diseases in a greenhouse, critical for early detection and mitigation strategies.
Carnegie Mellon researchers have shared a new in-home study with HAT (Head-Worn Assistive Teleoperation), a wearable device for people with severe motor impairments to teleoperate Stretch.
New research from RIKEN and the Nara Institute of Science and Technology explores training Stretch to take “reflective actions” in response to unclear human requests, using a crowdsourced dataset of real-world scenarios.
