Periodic walking motion of a Humanoid robot based on human data

Anne Kalouguine (1) , Christine Chevallereau (1, 2) , Sébastien Dalibard (3) , Yannick Aoustin (1)

1 LS2N - équipe ReV - Robotique Et Vivant
2 LS2N - Laboratoire des Sciences du Numérique de Nantes
3 SoftBank Robotics Europe

Human walking has been intensely studied, but it is di cult to reproduce on humanoid robots that maintain awkward movements. Three main di culties exist. (i) Di↵erent joint kinematics and size between humans and robots. (ii) A rolling motion of the foot which is often impossible to execute with humanoid robots that walk with their feet flat. (iii) A di↵erence in the dynamic model of a robot compared to a human that makes a copy of a human movement lead to unstable walking. In order to take into account the first two di culties, the specifications for reproducing human movements are adjusted. To ensure stability, a previously developed dynamic model called Essential Model is used. The zero moment point (ZMP) is imposed, and the horizontal evolution of the centre of mass (CoM) is computed to satisfy the ZMP.

Mots clés: Humanoid robot; Human-like walking; Center of Mass; Zero Moment point; Essential Model

Domaines: Robotique [cs.RO]

Anne Kalouguine, Christine Chevallereau, Sébastien Dalibard, Yannick Aoustin. Periodic walking motion of a Humanoid robot based on human data. EuCoMeS 2020 New Trends in Mechanism and Machine Science. EuCoMeS 2020. Mechanisms and Machine Science, pp.349-359, 2020. ⟨hal-03022032⟩