Zero-moment point feedback balance control of leg-wheel hybrid structures by using dynamic decoupling and control allocation

This paper proposes a balance control algorithm which uses Zero Moment Point (ZMP) in order to improve the mobility of leg-wheel hybrid structures on hard, flat surfaces. The algorithm generates balancing leg motions which continuously adjust the contact points to allow the geometric center of the contact area to follow ZMP. During our simulation of a balance control system which uses Individual Joint PD Control (IJPD), we encountered two main problems-high frequency oscillation and unbalanced control input. To solve these issues, we derived a method for dynamic decoupling and control allocation (DDCA). We tested the effectiveness of the proposed algorithm by simulating the maximum distance that a robot can stably move during a fixed time period. There are three important advantages to our proposed balance control algorithm. First, the balancing motions have no effect on the primary tasks of its arms and upper body. Second, the interaction forces between the contact points are minimized, thereby reducing an unbalance of the control inputs. Third, the task dynamics are decoupled and controlled independently.