SAGE: A Logical Agent-Based Environment Monitoring and Control System Krysia Broda 1⋆ , Keith Clark 1 , Rob Miller 2 , and Alessandra Russo 1 1 Department of Computing, Imperial College London www.doc.ic.ac.uk/{~kb/,~klc/,~ar3/} {k.broda,k.clark,a.russo}@imperial.ac.uk 2 Department of Information Studies, University College London www.ucl.ac.uk/infostudies/rob-miller/ rsm@ucl.ac.uk Abstract. We propose SAGE, an agent-based environment monitor- ing and control system based on computation logic. SAGE uses forward chaining deductive inference to map low level sensor data to high level events, multi-agent abductive reasoning to provide possible explanations for these events, and teleo-reactive programming to react to these expla- nations, e.g. to gather extra information to check abduced hypotheses. The system is embedded in a publish/subscribe architecture. Key words: Environmental Control, Logic, Event Calculus, Logic Pro- gramming, Abduction, Multi-Agent Reasoning, Teleo-Reactive Programs 1 Introduction SAGE (Sense, Abduce, Gather, Execute) is an agent based environment mon- itoring and control system we are building based on computational logic. Its key components are: (1) the use of the event calculus [7] and forward chain- ing deduction to map low-level time-stamped sensor reading events into inferred higher-level composite events using specialist agents for each type of sensor, (2) the use of multi-agent abductive reasoning [6] to co-operatively infer unobserved events or actions as possible explanations of the composite events, (3) the use of agents executing persistent teleo-reactive [9] control procedures that automat- ically respond to possible explanations, to gather auxiliary information about the state of the environment or to execute goal directed and robust control re- sponses, (4) the use of a formal logical ontology to specify application dependent event terms, facts and action descriptions, and (5) the use of publish/subscribe servers [12] as a communication infrastructure enabling easy integration of new components and agents that have only to conform to the application ontology regarding the notifications they publish and the subscriptions they lodge. The key advantage of using computational logic is that it allows high level declarative representation of an application domain. It greatly facilitates exten- sibility and re-engineering for different domains. Throughout this paper we will provide further commentary on the features listed above with reference to the following example scenario. Ann, Bob and Carl live in a sheltered housing complex which employs SAGE. At 9:30:00am and 9:30:09am respectively, adjacent sensors s34 and s35 detect movement down the main corridor of the complex. The building agent infers that someone is ⋆ Authors are listed in alphabetical order, not in order of their relative contribution. Published as a "short paper" in the proceedings of the 3rd European Conference on Ambient Intelligence (AmI'09), November 18th - 21st, 2009, Salzburg, Austria (see www.ami-09.org), published by Springer (LNCS series).