Learning in Agent-Based Concurrent Design and Manufacturing Systems

This chapter introduces general concepts, methods, and algorithms on multi-agent learning, with explanations and related examples on agent-based concurrent intelligent design and manufacturing systems. Concurrent Design and Manufacturing systems operate in real-time, dynamic environments. In an agent-based engineering design system, new computational tools or human specialists may have to be integrated into the working environment, and existing tools may breakdown or be removed from the running environment. A multi-agent design system includes many agents with many potential interactions. Learning is even more important in agent-based manufacturing systems. Learning in multi-agent systems is more than a mere magnification of learning in single agent systems. In agent-based concurrent design and manufacturing systems, the way that agents make information available to other agents determines where learning opportunities arise. Some multi-agent learning systems are implemented with training agents. In this case, the training agents or tutor agents are the main knowledge source of learning for the system.