Title :
Integration of sensorimotor mappings by making use of redundancies
Author :
Hemion, Nikolas J. ; Joublin, Frank ; Rohlfing, Katharina J.
Author_Institution :
CoR-Lab., Bielefeld Univ., Bielefeld, Germany
Abstract :
We present a novel approach to learn and combine multiple input to output mappings. Our system can employ the mappings to find solutions that satisfy multiple task constraints simultaneously. This is done by training a network for each mapping independently and maintaining all solutions to multivalued mappings. Redundancies are resolved online through dynamic competitions in neural fields. The performance of the approach is demonstrated in the example application of inverse kinematics learning. We show simulation results for the humanoid robot iCub where we trained two networks: One to learn the kinematics of the robot´s arm and one to learn which postures are close to joint limits. We show how our approach can be used to easily integrate multiple mappings that have been learned separately from each other. When multiple goals are given to the system, such as reaching for a target location and avoiding joint limits, it dynamically selects a solution that satisfies as many goals as possible.
Keywords :
humanoid robots; learning (artificial intelligence); robot kinematics; dynamic competitions; humanoid robot iCub; inverse kinematics learning; multiple input mappings; multiple mapping integration; multiple output mappings; multivalued mappings; neural fields; robot arm kinematics; sensorimotor mapping integration; Input variables; Joints; Manifolds; Neurons; Robot sensing systems; Training;
Conference_Titel :
Neural Networks (IJCNN), The 2012 International Joint Conference on
Conference_Location :
Brisbane, QLD
Print_ISBN :
978-1-4673-1488-6
Electronic_ISBN :
2161-4393
DOI :
10.1109/IJCNN.2012.6252487