In recent years, roboticists have developed a wide range of systems designed to tackle various real-world tasks, ranging from completing household chores to delivering packages or finding target objects in delineated environments.
A key objective in the field has been to develop algorithms that allow the reliable transfer of specific skills across robots with different bodies and characteristics, which would help to rapidly train robots on new tasks, broadening their capabilities.
Researchers at UC Berkeley have developed RoVi-Aug, a new computational framework designed to augment robotic data and facilitate the transfer of skills across different robots...
Read More
Recent Comments