HERB Masters Fine Manipulation Task with OREO Cookie Separation-Quality of Life Technology Center - Carnegie Mellon University

Monday, March 11, 2013

HERB Masters Fine Manipulation Task with OREO Cookie Separation

The Quality of Life Technology Center's Home Exploring Robot Butler, better known as HERB, is featured in a YouTube video that is part of OREO's ongoing "Cookie vs. Creme" campaign. The OREO twist challenge allowed researchers to validate manipulation algorithms that explore how robots can assist people at home, in ordinary kitchen environments, on simple cooking and snack preparation tasks.

HERB is a two-armed autonomous robot built to perform everyday manipulation tasks and instrumental activities of daily living (IADLs) with and around humans in cluttered and uncertain environments - like our homes and offices.

The video, shot Feb. 12 in the Personal Robotics Lab in Newell-Simon Hall, debuted March 8.  HERB twists an OREO apart and scrapes off what it terms "the precious creme" in the video, the fourth and final "Oreo Separator" machine in the online series.  Sidd Srinivasa, associate professor of robotics, and Pras Velagapudi, project scientist, play prominent roles as well.

"Cookie vs. Creme" was launched with the "Whisper Fight" commercial during this year's Superbowl broadcast. Three machines were custom-made for the humorous, online "Oreo Separator" series; HERB was the only research robot re-purposed for the videos.

Carnegie Mellon has produced a "behind the scenes" video regarding HERB's video shoot for OREO.

Srinivasa said the team had two weeks to prepare for the challenge; a team of four devoted about an hour a day to testing. Said Srinivasa:  "We accepted this challenge as an interesting research opportunity that could help validate our algorithms.  We were very pleased to see how well our algorithms actually performed."

Normally, HERB serves as a QoLTbots testbed system for the Quality of Life Technology Center, which is co-directed by Srinivasa. A complete mobile manipulation platform, HERB is equipped with perception, learning, and manipulation algorithms to discover and recognize objects lifelong, reconfigure cluttered scenes with physics-based actions like pushing and sweeping, and learn to produce safe, smooth, and intent-expressive motion for collaborative manipulation.


Learn More:

By: Byron Spice and Kristen Sabol, bspice@cs.cmu.edu, 412-268-9068