A Bipedal Controller Design to Realize Human Gaits

This research aims to mimic the exact gait of a human. In other words, we try to build a biped digital twin for human locomotion. The biped control has been a popular topic for years, and the various control methods are becoming matual. Advanced gait control for jumping, step climbing, and recovery from pushing have all been developed. Recent research has also addressed how to conduct a human-like gait. However, these results all used gaits that balance the biped instead of asking the biped to mimic the exact gait of a human. The concept can be helpful when one tries to examine if the gait of a human is normal. It can also be used to monitor the performance of an exoskeleton. The proposed digital twin analyzes the dynamic characteristics of the entire human body and determines the joint torques required for maintaining balance. This study constructed a bipedal robot model that conforms to the human body to simulate human postures. The zero moment point is used as a reference point for the interaction with the ground. We then used a cart table model to balance the bipedal robot. The controller outputs an assistive force at the center of mass position to maintain balance during walking. Finally, we convert the assistive force into ankle joint outputs to enhance gait stability and ensure that the waveform and numerical outputs are close to actual human data in the literature.