T he traditional centralized, sequential information-processing methodology can no longer meet the rapidly changing manufacturing environment's demands. An organization's information-processing capabilities are increasingly limiting its ability to quickly respond to changes and maintain productivity. Furthermore, classical control systems employ feedback techniques and models that generally can't manage computational complexity, nonlinearity, and uncertainty. 1 Classical control can't dynamically adapt well to the variability of the processes under control. However, agent-based control can easily handle combinatorial complexity in real time.Agent technology also provides an appropriate framework to integrate knowledge with production actions. 2-7 Knowledge integration depends on balanced information in and across organizations. So, replacing the central models with distributed intelligent agents will make it possible to cope with demands more efficiently. In this, each agent will represent a specific activity to support overall operations. With agents, maintaining the overall system is easier because instead of changes occurring to a colossal program, changes take place on a small scale without disrupting overall operations. However, building and testing such a distributed control system has been problematic at best.We've developed a set of tools and methodologies to help design, build, test, and verify such systems. Using these tools and methodologies, we've devised an agent architecture that we've applied to a US Navy shipboard chilled-water system. Simulation results indicate that our approach can lead to systems that operate in a real-life application with critical constraints such as survivability, dynamic reconfiguration, and reduced human supervision. We can apply the agent tools and methodology to develop agent-based solutions to a wide variety of industrial, financial, and homeland security problems.
Requirements for intelligent agent systemsTo be suitable for industrial automation, a collaborative intelligent system has at least four requirements:• Decentralization. Autonomous agents concurrently process and distribute information. • Goal centricity. Agents cooperate in and outside their domains to fulfill common objectives while satisfying constraints and optimizing the use of available resources. • Common knowledge and language. Agents use a domain ontology. • Scalability. Agents can either enter or depart a collaborative session according to need.
Agent collaboration infrastructure to enable intelligent behaviorIntelligent agents' core requirement is to behave intelligently. These agents are software entities and therefore can be programmed in many different ways to mimic the human brain. However, to achieve advanced reasoning in artificial systems requires sophisticated, expensive computing units. To achieve AI's ultimate goal will require completing a considerable amount of research on reduced-size computing units and, most essentially, small inference engines.Because we're confined to small units ...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.