A new ETH Zurich system enables robots to perform tasks with minimal human guidance

Two main categories of the system used are object-centric and robot-centric.

Researchers from ETH Zurich have developed a system for robots to perform tasks with minimal human guidance. The system follows a three-step process, beginning with the user describing the scene and action, followed by the system planning a route, and finally refining the route into a minimum viable path. Two main categories of the system used are object-centric and robot-centric. The object-centric relates to tasks like opening a door or dishwasher, while the robot-centric focuses on moving the robot around objects. The system is believed to use task specifications and high-level descriptions of the robot and object along with a sparse objective to discover how the robot should move, what forces it should exert, which limbs it should use, as well as when and where it should establish or break contact with the object.

The researchers suggest that the system could have applications in sectors such as manufacturing and healthcare. However, the system still has limitations that need to be addressed, including the robots’ data acquisition, environmental uncertainty, and physical constraints limiting range of motion. Further research is needed to make the system more reliable and efficient. Although robots have been able to perform tasks for some years now, they often require some human guidance. The new system is a significant step in reducing human input levels, which could lead to increased efficiency and effectiveness in a range of industries. The system may ultimately help address labor shortages for certain tasks and roles.