Robots Observe Humans to Learn Household Tasks

Researchers at Carnegie Mellon University developed the In-the-Wild Human Imitating Robot Learning (WHIRL) algorithm to teach robots to perform tasks by observing people. WHIRL enables robots to gain knowledge from human-interaction videos and apply that data to new tasks, making them well-suited to learning household chores. The researchers outfitted a robot with a camera and the algorithm, and it learned to complete more than 20 tasks in natural environments.

In each case, the robot watched a human execute the task once, then practiced and learned to complete the task by itself. Instead of waiting for robots to be programmed or trained to successfully complete different tasks before deploying them into people's homes, this technology allows us to deploy the robots and have them learn how to complete tasks, all the while adapting to their environments and improving solely by watching.

More information:

https://www.scs.cmu.edu/news/2022/whirl-robots

Robot Better than Surgeon in Precision Training Task

A robot outperformed a seasoned surgeon in completing a common training task with equal precision during a study conducted by a multinational team of researchers. Researchers at South Korea's Daegu Gyeongbuk Institute of Science and Technology tracked the arm movements of a da Vinci robotic surgical assistant using 3D printed markers.

Analysis via a machine-learning algorithm suggested the trained model can reduce the mean tracking error by 78%, from 2.96 millimeters (mm) to 0.65 mm. An experienced surgeon and nine volunteers with no surgical experience used a da Vinci system to complete three variations of a peg transfer task. The fully automated robot performed the bilateral tasks faster and more accurately than the surgeon.

More information:

https://spectrum.ieee.org/robot-outperforms-a-surgeon-in-a-precision-training-task

Google’s Prototype AR Glasses

Google will be testing a small number of prototypes within select areas of the US with strict limitations on where testers can operate, and the kinds of activities they can engage in. Testers will have to go through device, protocol, privacy, and safety training. And the company is warning that it will have prototypes that look like normal glasses, though they’ll have an in-lens display and visual and audio sensors like a microphone and camera onboard. An LED indicator tells people in the vicinity if image data is being saved for analysis and debugging, which they can request to have deleted.

Google is planning to explore use cases like speech transcription and translation, as well as visual sensing scenarios like translating text or helping with navigation. The company claims that its prototypes don’t support photography or videography, though any image data captured during its tests will be deleted unless the data is used for further analysis or debugging. In that case, the image data is first scrubbed for sensitive content, including faces and license plates. Then it is stored on a secure server, with limited access by a small number of Googlers for analysis and debugging and deleted after 30 days.

More information:

https://www.theverge.com/2022/7/19/23270219/google-ar-prototypes-test-public