Human AugmentatioN via Dexterity
Revolutionizing Robot Dexterity and Empowering Human Work
Defining Dexterous Skills
Defining Dexterous Skills will develop a general framework for representing dexterous robot manipulation skills. We will identify common components of skills and how they incorporate different sensor modalities as inputs. We will explore how human input can be used to adapt skills, and how skills can be combined into a hierarchy to accomplish complex tasks. The framework will allow us to analyze and compare different types of skills, and it will form the basis for our dexterous skill library.
Project Lead

Oliver Kroemer
Carnegie Mellon University
Project Team

Gary Fedder
Co-lead, TB2-Assembly
Carnegie Mellon University

Rebecca Friesen
Texas A&M University

Roberta Klatzky
Carnegie Mellon University

Kevin Lynch
Northwestern University

Melisa Orta-Martinez
Carnegie Mellon University

Julie Shah
Massachusetts Institute of Technology

The picture shows a variety of example manipulation skills from previous works that we aim to capture using a unified skill framework.