This paper introduces a motion planning method for a rapid mobile manipulator using an inverted pendulum model. We design a linear quadratic optimal controller to stabilize ZMP. The two kinds of ZMP stabilization strategies (Fixed ZMP and Relaxed ZMP) are proposed. The highly geared manipulator is controlled by a Cartesian computed torque (CCT) control for compliant motion. A rapid mobile manipulator called KDMR-1 has been developed for its application. The high acceleration and speed performances of the proposed methods are shown by a rapid maneuvering experiment.

In this paper, we developed a novel dynamic mobile platform capable of transforming between four wheel rapid driving mode with ZMP stabilization and two wheel balancing mode. Usually, the four wheel mobile platforms have low acceleration due to the high center of mass. Using the feed-forward ZMP stabilization method, we achieved high acceleration performance with maintaining stable ZMP. We use a LQR controller for two wheel balancing mode. In the two wheel balancing mode, there are two different driving modes, self-balancing mode, human-riding mode. The transformation between four wheel and two wheel mode is made possible by swing up and swing down control method. We prove the dynamic performance of mobile platform by experimental results.

This paper introduces a novel approach to control a wheeled inverted pendulum when a disturbance is applied by a human. The interaction between a human and a wheeled inverted pendulum involves a pulling or pushing force. This type of action is a severe disturbance for a wheeled inverted pendulum, as the wheeled inverted pendulum tends to maintain its initial position if there is no desired input. Thus, there are many possibilities for the wheeled inverted pendulum to be unstable as a result of interactions with humans. To solve this problem, the control algorithm of a wheeled inverted pendulum was designed to move in coordination with the external force of a human. This control algorithm is termed human-friendly motion control. It contains an optimal controller using a full-state feedback and a reduce-order disturbance observer. The disturbance torque from a human was estimated, and the estimated disturbance torque was used to generate a position reference for the human-friendly motion. This control algorithm keeps the wheeled inverted pendulum stable even when a severe disturbance is applied.