Human-inspired walking-style for a low-cost biped prototype

Abstract

Many ongoing projects are being done in the field of humanoid robotics. Studies in the generation of human-like motions for humanoids based on human-motion-captured-data (HMCD) have demonstrated that this is feasible. Walking style patterns using only the ZMP concept have shown that the resulting motion of humanoids are not natural as human's motion. By combining human-motion captured-data with other techniques, it is possible to improve humanoids' walking style. On this matter, this work is intended to apply HMCD to MAYRA humanoid prototype and show its advantages versus the classic statically stable humanoid walking with semi-circular foot-trajectory. We conclude that signals obtained directly from human-walk are smoother rather than those signals obtained from mathematical foot-trajectory. This may lead to a simplification in the control system for the MAYRA biped prototype