ID offerECND39
UniversityEcole Centrale de Nantes
EU Subject SubCode06.1
Title of the course/domaine/keywordOnline whole body human motion conversion for humanoid robot control using balance constraints
Explanatory textCurrent approaches for human imitation by humanoid systems are mainly offline and do not allow a direct real-time motion conversion. The main problem is to deal with the balance constraint while moving at the same velocity as the human model. In our research group, an online kinematic approach was developed and validated experimentally , showing good results for slow movements as the dynamics effects for high accelerations were not modeled and were very different for the natural and the artificial systems. Other offline studies allow dynamics conversion of the motion for point to point motion or cyclic walking motion. This thesis topic will focus on enlarging the models to online dynamics motion conversion in order to increase the admissible motion velocity for real-time imitation. The experimental protocol will use the available 8-camera optical motion capture to acquire the human motion, and the Aldebaran humanoid robots NAO (small size) then ROMEO (mid-size). Once the experimental environment is set, the candidate will focus on an evaluation criterion of the motion imitation and propose a learning process to improve the characterization of the motion patterns for the studied human subjects.
Max number of scholarships1
Maximum Duration (months)36
Earliest start dateJuly 1, 2013
Latest end dateJuly 14, 2016
Contact NameChristine Chevallereau, Sophie Sakka
Possibility to learn the national languageYES
Language Requirements
Academic Requirementsmasters
Additional information