Movement recognition and cooperative task synthesis through hierarchical database search

Abstract

An approach for movement recognition and cooperative task synthesis is presented. It is based on a two-part hierarchical database, one consisting of human motion and the second on cooperative robot movements. While the motion recognition is done through hierarchical search on the primary part of the database, the secondary part is used for determining the most probable path for cooperative movement synthesis. Dynamic movement primitives are used to encode the path into a smooth and continuous movement. Initial evaluation, done in simulation, shows the validity of the proposed approach.