Atomic Actions for Home-Service Robots

Abstract

In ubiquitous computing environments, home-service robots are expected to perform many kinds of tasks. It is difficult, however, to write programs for all the tasks in advance because of the diversity of the tasks and changing environmental conditions. One of the essential processes of autonomous multi-functional robots is to define a set of basic robot actions that can be executed unambiguously and also checked for completion. A task planner then can use these actions to accomplish complex tasks that home-service robots are expected to do. This thesis first proposes a set of tasks for the first-generation home-service robots, and then systematically decomposes them into sequences of smaller but meaningful actions called molecular actions. The molecular actions are then decomposed into yet more primitive actions called atomic actions. Because vision, sound, range sensors, and force sensors are the primary means of monitoring task progress and completion, the atomic actions are classified according to the complexities and frequencies of the sensing algorithms used. The resulting taxonomy of atomic actions serves as basic building blocks for a knowledge-based task planner. In addition, the representative atomic actions have been implemented on a real home-service robot for justifying the advantages of using the actions.