Interaction mesh based motion adaptation for biped humanoid robots

S. Nakaoka, T. Komura

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Adapting human motion data for humanoid robots can be an efficient way to let them conduct specific tasks or perform visually intriguing movements. However adapting movements which involve close interactions between body parts / objects has been a difficult problem if we try to make the motions sufficiently similar to the original ones without causing undesired collisions and fall-overs. To solve this problem, we propose a novel motion adaptation system which can adapt full-body motions of a human to those by a biped humanoid robot. The system integrates the joint coordinate model and a dynamic balancing method with the interaction mesh based motion adaptation method, which is originally designed for character motions represented by marker positions in Cartesian coordinate.We present that the system can adapt human motion sequences that include close interactions between the arms for the biped humanoid robot HRP-4C, and the adapted motion can be stably performed in dynamics simulation.
Original languageEnglish
Title of host publicationHumanoid Robots (Humanoids), 2012 12th IEEE-RAS International Conference on
Number of pages7
Publication statusPublished - 1 Nov 2012


  • control engineering computing
  • human-robot interaction
  • humanoid robots
  • legged locomotion
  • motion control
  • robot dynamics
  • Cartesian coordinate
  • adapted motion
  • biped humanoid robot HRP-4C
  • biped humanoid robots
  • character motions
  • dynamic balancing method
  • dynamics simulation
  • full-body motions
  • human motion data
  • human motion sequences
  • interaction mesh based motion adaptation method
  • joint coordinate model
  • marker positions
  • motion adaptation system
  • visually intriguing movements
  • Adaptation models
  • Dynamics
  • Foot
  • Humanoid robots
  • Joints
  • Trajectory


Dive into the research topics of 'Interaction mesh based motion adaptation for biped humanoid robots'. Together they form a unique fingerprint.

Cite this