A semantic constraint-based Robot Motion Control for Generalizing Everyday Manipulation Actions
Ingo Kresse · 2017
In their everyday activities, humans implicitly know what is important in their movements and how they can adapt them to changes in their environment, like changed object locations, clutter or different furniture. This thesis endows robots with some of this task knowledge and introduces a modular constraint-based movement description that combines geometric features like edges or corners and abstract terms like alignment, distance or contact force and turns them into executable movements.