Communications of the ACM
Robot Learns to Clean Space Just the Way You Like It
TidyBot is trained within a specific space so it can learn how different types of objects are put away in that area.
Credit: Kurt Hickman
Roboticists at Stanford University, Princeton University, The Nueva School, Columbia University, and Google collectively developed a robot that can clean spaces based on users' personal preferences.
The TidyBot is equipped with wheels under a box-shaped platform, as well as a seven-jointed arm and a two-finger gripper, allowing it to navigate spaces while picking up objects to be placed in certain areas or tossed in the trash.
It can identify objects and obstacles via ceiling-mounted cameras, and view and analyze items close up using cameras attached to its body.
The researchers trained TidyBot to identify broad categories of objects using a large language model, meaning users can provide examples about their preferred locations for particular items and have the robot understand how to organize similar items it may never have seen.
However, the researchers said TidyBot currently has trouble with items that are hard to grasp, like a credit card resting on a table, and with objects placed in new locations.
From Stanford News
View Full Article
Abstracts Copyright © 2023 SmithBucklin, Washington, D.C., USA
No entries found