Home | english  | Impressum | Sitemap | KIT

Whole-body action learning and planning with affordances

Whole-body action learning and planning with affordances
Forschungsthema:Whole-body action learning and planning with affordances
Typ:Bachelor/Master Thesis

Peter Kaiser 

Whithin the European research project WALK-MAN we aim at enabling our humanoid robot ARMAR-IV to autonomously operate in damaged or hazardous sites like destroyed buildings or power plants.

We plan to apply the cognition psychological concept of affordances to whole-body manipulation and locomotion. The concept of affordances states that objects or parts of objects afford possible actions to the agent. A cup for example affords grasping and drinking, a chair affords sitting. Recognizing and utilizing available affordances is a crucial capability for robots that operate autonomously in unstructured environments. We are especially interested in affordances that affect the whole body of the robot, for example the utilization of handles or handrails for enhanced stability while walking.

Possible topics for bachelor or master theses are:

  • Representation of affordances within the terminology of Object-Action-Complexes (see Xperience)
  • Recognition of affordances based on visually detected primitives
  • Utilization of affordances for stable loco-manipulation
  • Planning with affordances

Applicants should be proficient in programming, preferably in C++.