Towards agent-oriented smart office based on concurrent logic languages

Abstract

This paper presents an AI architecture for multiple robots working collaboratively in a future smart office. This architecture integrates the control, communication, planning, and learning necessary to agentify office robots. Such integration is based on our multiagent robot language (MRL), which is an extension of concurrent logic programming languages (CCL). While the behavior of an agent is specified within guarded Horn clause logic, the communication and concurrency controls are amenable to the operational semantics of CCL. Our planning module provides well-balanced coordination between single agent planning and supervisor-level planning, yielding the minimum interaction for multiagent communication and control. Furthermore, inductive learning is incorporated into the planning module and is applied to produce empirical rules for action selection, providing the utility of multiagent problem solving. These features allow a unified view of both low- and high-level computation, which enables intelligent collaboration between robotic agents and provides a powerful framework for distributed AI and agent-oriented programming in the real world.