Action selection in teleautonomous systems

Abstract

Numerous generalized architectures for telerobotic systems have been proposed. Most are hard-wired instantiations of a specific approach, which rarely address the interactions between human and autonomous elements of teleautonomous systems. The work described here provides a more general, easily extensible and modifiable architecture. Specific issues addressed include methods that fuse decisions from multiple sources (such as human, reactive, or deliberative), integrate multiple command sources, and accommodate multipurpose event recognizers to change operating modes. These methods may be dynamically altered to generate a variety of control modes. A result of the research described in this paper is a modifiable, extensible framework for the unification of the many modes of control, such as shared, teleoperative, and supervisory, which are described in the telerobotics literature.