Cognitive map architecture

Abstract

We have developed the Cognitive Map robot architecture that minimizes the amount of rewriting of existing legacy software for integration. The Cognitive Map can be thought of as a centralized information space for connected components to contribute both internal and environmental state information. We leverage several successfully proven concepts such as blackboard architectures and publish- subscribe based messaging to develop a flexible robot architecture that exhibits fault-tolerance, easily substituted components, and provides support for different structural paradigms such as subsumption, sense-plan-act and three-tier architectures. Our multicomponent distributed system has system components that are loosely coupled via message-passing and/or continuous data streams. This architecture was implemented on the humanoid robot ASIMO manufactured by Honda Motor Co., Ltd. We review various forms of communication middleware and component models. The "Architecture" section provides an overview of our architecture and considerations in its design. The "Scenario Design" section details the process from conceptualizing an interactive application to its instantiation in the robot architecture. The "Components" section singles out several important high-level components that play a significant role in many of our interactive scenarios. Finally, discussions and conclusions are presented.