バケットマニピュレーションの軌道生成と制御 ホイールローダの知能化に関する研究 ＩＩ

Abstract

In order to provide capability of intelligent control to the wheel loader, a new method for generating trajectory and controlling the bucket manipulator during scooping is proposed. In a strategic point of view, the scooping process is divided into three phases, which take place in sequence. The bucket is pushed into the piled material in parallel with the ground surface in the first phase, and then it is moved upward and forward simultaneously in the second phase. In the final phase, bucket is lifted away from the piled material. The control target and variables are designated differently in each phase. In this study, a trajectory along which the bucket tip moves is extended step by step in a regular time interval, instead of generating whole trajectory in advance before the start of each phase. The new method of trajectory generation can adapt various slope geometry of piled material without any information of the pile in advance. Fuzzy reasoning is also adopted to generate trajectories of the three actuators consisting of the bucket manipulator. The error of the current trajectory from the target that might be caused by disturbances is compensated by feed forward control. A new scheme combined fuzzy incremental generation of trajectory with feed forward control is validated by a series of computer simulation which covers scooping of nine different slope conditions of piled material.