Experimental setup: simulated (top) and real (bottom) robotic environments and tasks. Credit: arXiv (2024). DOI: 10.48550/arxiv.2409.06613
Engineers working on Google’s DeepMind project have announced the development of two new robotic systems powered by artificial intelligence. One, called ALOHA Unleashed, was developed to advance the science of two-armed manipulation. The other, called DemoStart, was developed to advance the capabilities of robot hands with multiple fingers, joints or sensors.
Details of ALOHA Unleashed have been published on the DeepMind website and also on GitHub. Details of DemoStart have been published on the arXiv preprint server.
As the research team points out, most robot hands designed to grasp and move objects typically act alone, without a second hand to assist them. In this new project, the research team used AI technology to teach a robot to use both hands at the same time to accomplish a “difficult” task, such as tying a shoe. The result is ALOHA Unleashed.
As the team also points out, the new system builds on ALOHA 2 and the ALOHA platform, developed at Stanford University for use in teleoperation applications. The new system improves dexterity and also allows two robot hands to become “aware” of each other when working together on a common problem.
The robot hands were taught by demonstration how to perform tasks such as hanging a shirt or repairing a part of the robot. Then, diffusion methods were applied to give the robot hands a degree of prediction, helping them anticipate what the other would do.
The DemoStart research team found that complex robot dexterity would require more fingers, joints and sensors than are currently used on most robot hands. To achieve this, a certain degree of coordination between them is required.
As in the ALOHA Unleashed project, coordination required the introduction of AI into the learning process. With DemoStart, they used reinforcement learning to help the robot gain a sense of its capabilities when given control of multiple joints in the arms, hands, and fingers, in addition to its fingertips.
The approach was to give the robot’s hands simple tasks and gradually increase the difficulty. They found that they could teach a two-fingered robot with multiple joints and sensors to reorient a cube, tighten a nut, and tidy up a workspace.
More information:
Maria Bauza et al, DemoStart: Demonstration auto-program applied to real-time simulation with multi-fingered robots, arXiv (2024). DOI: 10.48550/arxiv.2409.06613
ALOHA Unleashed: A Simple Recipe for Robot Dexterity, aloha-unleashed.github.io/asse … /aloha_unleashed.pdf
DeepMind Blog: deepmind.google/discover/blog/ … -in-robot-dexterity/
DemoStart: sites.google.com/view/demostart
arXiv
© 2024 Science X Network
Quote: Google DeepMind unveils two new AI-powered robotic hand systems: ALOHA Unleashed and DemoStart (2024, September 23) retrieved September 23, 2024 from
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.
