Planning and control are two basic components of autonomous robot systems. In recent years, significant advances have been made in both of these two areas. However, these developments have continued to be carried out in isolation, and the gap between planning and control remains an important one. The discussion will focus on our ongoing effort for the extension of real-time robot control capabilities in the context of integrated robot planning and control. This effort is directed at the development of high-level real-time connections between perception and action. The goal is to render robot operations less sensitive to tolerances and uncertainties, allowing robust execution when the environment is not known exactly. The aim of this effort is the development of a layer of real-time sensor-based motion primitives for both gross motion and fine motion operations. Central to the development of such capabilities is the design of control strategies that can deal with the robot task-level behavior. Another important issue in the development of such capabilities is the integration of these real-time control strategies with global task planning algorithms. Examples of appplication areas and experimental results associated with the implementation of the various methodologies developed at Robotics Laboratory will be presented.
This tutorial covers the following fundamental concepts and technologies:
The WIDE (Widely Integrated Distributed Environment) project
is a consortium of more than 200 researchers on Internet area
which covers the development of above technologies.
The works by the WIDE project will be described as case studies.