CMU researchers teach robots to do chores by watching humans

-
WHIRL robot

Researchers at Carnegie Mellon University (CMU) have developed a learning method for robots that allows them to perform household tasks after watching a person do them just once. | Source: Carnegie Mellon University

A research team from Carnegie Mellon University’s (CMU) School of Computer Science has developed a new method for robots to learn called WHIRL, which stands for In-the-Wild Human Imitating Robot Learning. 

WHIRL is an efficient algorithm for one shot visual imitation. With WHIRL, a robot can learn to perform household tasks just watching a person perform them. 

The CMU team added a camera and its algorithm to an off-the-shelf robot to test the abilities of its software. When testing the robot, the team found that it was able to perform over 20 tasks after watching someone perform them just one time. 

The robot learned how to do things like opening and closing appliances, cabinets, doors and drawers, putting a lid on a pot, pushing in a chair and taking the trash bag out of the can, among other things. None of the items the robot interacted with, whether it be appliances or doors, were modified to suit the robot. 

While the robot’s first few attempts at most tasks failed, it was able to quickly latch onto how to perform the task correctly after a few successes. 

The robot often completed tasks using different movements than the humans who demonstrated them, but the Carnegie Mellon team isn’t concerned about that. WHIRL doesn’t aim to make a robot, equipped with different tools than a human, perform a task the same way a person would. Instead, the robot focuses on getting to the same end result. 

Robots typically learn to do tasks with one of two methods. The first, called imitation learning, involves humans manually operating a robot to teach it a task. The second, called reinforcement learning, requires robots to learn from millions of examples in simulation, then adapting that training to the real world.

Both methods involve repeatedly teaching the robot to perform a task, making it difficult to teach a robot to perform multiple tasks. With WHIRL, a robot can learn multiple tasks quickly, with a human only having to demonstrate it once. 

Shikhar Bahl, a PhD student at Carnegie Mellon’s Robotics Institute (RI), worked with Deepak Pathak and Abhinav Gupta, faculty members at the RI on the research. The team presented WHIRL at the Robotics: Science and Systems conference in New York earlier this month. 

The post CMU researchers teach robots to do chores by watching humans appeared first on The Robot Report.



from The Robot Report - Robotics News, Analysis & Research https://ift.tt/Oqtn62P
via artificialconference

Comments

Post a Comment

Popular posts from this blog

Valiant TMS and Realtime Robotics partner to cut programming, cycle times

-
Image
Valiant TMS is integrating robot motion optimization as a service. Source: Realtime Robotics Realtime Robotics today announced that Valiant TMS will integrate its Optimization-as-a-Service offering with customers’ manufacturing operations. The companies said the combination of Realtime’s proprietary software and engineering expertise with Valiant TMS’s manufacturing industry experience will improve the efficiency and productivity of automated operations. With Optimization-as-a-Service, companies can identify and eliminate production bottlenecks, significantly reducing cycle time and programming time, said Realtime Robotics. “Manufacturers are always looking for ways to improve their production processes, but few have the time or resources to spend on such a lengthy, involved task,” said Peter Howard, CEO of Realtime Robotics, in a release . “We’re thrilled that our optimization solution has struck a chord with so many, automating the task of generating and testing potential layou...
Read more

Ocado wins UK patent lawsuit over Autostore

-
Image
On the left, Autostore’s ASRS, and on the right, Ocado’s ASRS. | Source: Autostore, Ocado Yesterday, the UK High Court ruled that Autostore’s patents, which have been the subject of a patent infringement case against Ocado , were invalid, and that Ocado did not infringe upon them. Both Autostore and Ocado operate automated storage and retrieval systems (ASRS) that operate on similar principles. Autostore filed the lawsuit in October 2020, when it asserted that Ocado infringed upon six of its patents. The European Patent Office invalidated two of these patents before judgment was made on the case, and two other patents were withdrawn by Autostore before the hearing started.  Judge Hacon invalidated the remaining two patents because AutoStore had disclosed its technology publically in a business deal with the Central Bank of Russia before the company filed its patents. Halcon ruled that even if the patents were valid, the Ocado Smart Platform (OSP) did not infringe on the...
Read more

Update on AgTech automation at CNHI

-
Image
The Robot Report Podcast · Update on AgTech automation at CNHI The Robot Report Podcast · 2023 year in review: GenAI, humanoids dominate robotics industry   In Episode 138 of The Robot Report Podcast, co-hosts Mike Oitzman and Eugene Demaitre catch up with Marc Kermisch, global chief digital and information officer at CNH Industrial, or CNHI. Kermisch takes us through the latest developments at CNHI in agriculture autonomy and digitization, explaining how robots and data can help farmers and growers improve yields and efficiency. We also recap the latest robotics news and walk through an update from robotics and automation companies that attended this month’s 2024 CES show. News of the week Locus Robotics layoffs The mobile robot maker last week confirmed that it had a “small, targeted RIF,” or reduction in force. North American robot orders dropped last year, according to the Association for Advancing Automation ( A3 ). Warehouse construction also declined by 2...
Read more