*Visiting Faculty, Innovative School of Business Management, Ahmedabad, Gujarat, India
**Research Engineer, Solusoft Technologies Pvt. Ltd., Ahmedabad, Gujarat, India
Intelligent agent systems are becoming popular and more important in our day-to-day activities and users are expecting more intelligent services from them. The paper proposes a third level of sophistication agent capable of flexible autonomous behavior. It provides volunteer information or services to a user, without being asked, whenever it is deemed to be appropriate. A prototypical agent-based refrigerator system has been described, where the environment is challenging; more specifically, it is dynamic, unpredictable and unreliable. The refrigerator agent-‘Janitoron’ with high complexity is expected to have all the characteristics in the ideal situation ranging from autonomy, informativity and both reactivity and pro-activity. The proposed agent maintains a balance between reactivity and pro-activity. An agent senses changes in its environment and responds to these changes. On one hand, the agent‘s actions are influenced by environmental changes and on the other hand, the agent‘s plans and actions are influenced by its goals. Being proactive and reactive, Janitoron behaves more like humans in the way it deals with different situations.
The paper describes the basic working of the refrigerator agent. The agent receives a percept containing information about the food kept at certain location in the refrigerator. Since the agent has no existing knowledge about the food, this percept results in updating of the agent‘s beliefs with new knowledge. This in turn generates a goal to take actions, which then leads to a plan being selected and executed. This way the agent can monitor status of food kept in refrigerator if the user don‘t have time do so.
Agent Specifications, Dynamic Environment, Intelligent agent, Percept Sequence, Refrigerator Agent