Real-time ZMP Compensation Method using Null Motion for Mobile Manipulators Jinhyun Kim Wan Kyun Chung Robotics & Bio-Mechatronics Lab., Pohang University of Science & Technology (POSTECH) San 31, Hyoja-Dong, Nam-Gu, Pohang, 790-784, Korea E-mail : {pluto,wkchung}@postech.ac.kr Abstract This paper presents a method to deal the dynamic stability for a mobile manipulator. Although the system has static stability, manipulation on the moving base or mobile locomotion with a manip- ulator may cause the system to turnover due to dynamics, so the controller of a mobile manipulator is carefully designed. In this paper, to define the dynamic stability for a mobile manipulator, Zero Moment Point (ZMP) is used. ZMP is very useful measure for the dynamic stability. However, if the degrees of freedom of the system are large, the calculation algorithm is very complicated. So, to simplify the calculation algorithm, we define ZMP using iterative Newton-Euler formulation. Next, a unified approach for the two subsystems, mobile and manipulator, is formulated using redundant scheme. And, to conserve the dynamic stability of the system in real-time, we define the perfor- mance index for the redundant system using ZMP. And then, the redundancy resolution problem for optimizing the proposed performance index is solved using the null motion optimization. Finally, the performance of the proposed method is demonstrated by simulation. keywords: Mobile manipulator, ZMP, redundant robot, null motion, dynamic stability 1 Introduction In recent years, a lot of practical tasks require manipulator motion under moving base frame, because of the large work area due to mobile platform, for example, service robotic application, inspection and repairing in the unmanned arena, etc. Also, via moving platform, the system has redundant degrees of freedom in end-effector, therefore, for the same end-effector motion, there exist infinite configurations of robot. On the contrary, different dynamic characteristics of the two subsystems make the control problem of whole system difficult, and the interaction forces between the subsystems give another control hardship. Until now, for the mobile manipulator systems, many researchers have studied on various topic, for example: optimal problems such as determining manipulator configuration and obstacle avoidance, 1