抄録
Offer Organization: Japan Society for the Promotion of Science, System Name: Grants-in-Aid for Scientific Research, Category: Grant-in-Aid for Scientific Research (C), Fund Type: -, Overall Grant Amount: - (direct: 3600000, indirect: 450000)
In former researches in mufti-robotics, in spite of its importance, there have been no propositions of task allocation scheme toward the problem involving "task constraints" that impose some restriction on the orders of task execution. In other studies involving task constraints such as job shop scheduling, task constraints are given by the designer. We address the problem that robots should calculate task constraints by themselves and should execute task allocation considering their constraints. As an example of such problem, we adopted rearrangement task of multiple movable objects. In this research, following subjects should be solved.
1. Task allocation should be executed considering task constraints.
2. Task constraints should be calculated as early as possible.
Toward these subjects, we have obtained following results.
1. By means of classification of task constraints, we have obtained calculation methods of task constraints and task allocation methods considering these constraints.
2. We have promoted efficiency of calculation procedure of task constraints. First, we have analyzed calculation procedure from viewpoints when calculation should be done and how long it should be taken. Next, we have listed all possible procedure. Finally, we have tested these procedures in simulated environments. As a result, following method turns out to be the most effective: In every time of task allocation, robots should calculate some constraints that costs comparatively low calculation costs. And robots should calculate other constraints that cost comparatively high calculation costs only when robots fail to planned tasks.
In future works, more flexible framework that divides given task into several small tasks is needed. For example, some objects should be temporarily set to adequate intermediate goals.