A Hybrid CPG–ZMP Controller for the Real-Time Balance of a Simulated Flexible Spine Humanoid Robot

摘要: Although a few researchers have started to realize the importance of flexible spine in generating more natural-looking behaviors for humanoid robots, there has been no work on trying make full-body robot (that spine) maintain balance. The main reason this is that it very costly and difficult develop control kind robot. This paper serves two purposes. First, proposes use realistic 3-D simulator as platform robotics research. Second, presents hybrid CPG-ZMP controller simulated biologically inspired CPG component our allows mechanical feet exhibit rhythmic motions using only parameters. Through monitoring measured ZMP location, engineering modulates neural activity allow balance while standing exhibiting sagittal frontal planes real time. final postures emerge automatically time through dynamic interactions between networks, itself, its environment. Since experimental results also demonstrated system robust against disturbances from external pushing forces, potential be applicable next generation robots.