Development of a rapid mobile manipulator and model-based stabilization methods

Abstract

This thesis describes about ZMP stabilization methods of mobile manipulators for rapid object delivery and model-based stabilization methods based on the compliant control of the manipulator. The wheeled mobile manipulator has the best mobility on the flat ground. However existing mobile manipulators have low acceleration and speed performances. In the case of mobile service robots, they have the certain level of high center of mass to interact with human. In that case, mobile service robots have a possibility to easily rollover in the high acceleration. To overcome this limitation, various researches about the motion planning method to stabilize the mobile manipulator have been studied. However, most of researches were confined to simulation results and did not apply to the actual mobile robot. This was due to the complexity of existing algorithms and the lack of the rapid mobile manipulator. In this thesis, we develop the rapid mobile manipulator and ZMP stabilization methods which are simply applicable to the actual mobile manipulator in the real-time. We prove the maximum velocity of over 12km/hr and the maximum acceleration of over 0.5g through experimental results. The flat ground which the rapid mobile manipulator moves has some irregularity. At this high speed, the ground disturbance due to the ground irregularity becomes a crucial problem for stability. To solve this problem, we propose the active suspension system equipped with the complaint control. The manipulator is controlled by the Cartesian computed torque method for compliance. Using these algorithms, we show that the object delivery is possible in a high speed though experimental results. The conversion algorithm to transform from a four-wheel driving mode and a two-wheel driving mode and vice versa is proposed. Though the conversion algorithm, it is possible to use a four-wheeled ZMP stabilization mode and a two-wheeled self-balancing mode. And also we propose the human-riding mode that the ...