A Task Planning System of Steward Robot for Human-friendly Human-Robot Interaction in an Smart Home Environment

摘要

This paper presents a task planning system for Steward Robot, which is an interactive intermediate agent between inhabitant and a complex smart home environment called the ISH (Intelligent Sweet Home) in KAIST. The ISH is a large-scale robotic environment with various assistive robots for the independent living of the elderly and the people with disabilities. In the ISH, as an approach toward human-friendly human-robot interaction, we aim 'simplification of task commands' by the user. In this sense, a task planning system has been proposed to generate an effective sequence of actions for coordinating sub tasks of the target subsystems from the given high-level task command. Basically, the task planning is performed under the framework of the STRIPS (Stanford Research Institute Problem Solver) representation and the splitted planning method. Note that one of the satisfaction measures, in view of performance, relies on the computation time of a planning system. And the computation time, in task planning, mainly depends on the number of predefined actions and facts in a world model. In this paper, we propose a task planning system for reducing computation time in the planning process by a state partitioning technique. First, the planning system constructs a backward splitted planning graph. Then, by analyzing the graph, it decomposes an original planning problem into a set of the independent sub-problems. Finally, the planning system generates a sequence of actions for each sub-problem using the backward splitted planning method. As a case study, the experimental result with a scenario-based task planning problem in the ISH is given to show the effectiveness of the proposed system.