33 DoF Quasi-Natural Humanoid Robot

The objective of this research is to develop a quasi-natural humanoid walking trajectory generator based on five-mass with angular momentum model using feedforward-feedback control. This approach aims to minimize the modeling error and improve the frequency characteristics due to non-minimum phase property, so that the tracking accuracy increases and walking performance can be enhanced. Using feedback and feedforward controller based on the proposed model, the humanoid walking pattern generator is verified and demonstrated with humanoid robot developed in our laboratory.