Information System for Storage, Management, and Usage for Embodied Intelligent Systems

Abstract

Embodied intelligent agents that are equipped with sensors and actuators have unique characteristics and requirements regarding the storage, management, and usage of information. The goal is to perform intentional activities, within the perception-action loop of the agent, based on the information acquired from its senses, background knowledge, naive physics knowledge, etc. The challenge is to integrate many different types of information required for competent and intelligent decision-making into a coherent information system. In this chapter, we will describe a conceptual framework in which such information system can be represented and talked about. We will provide an overview about the different types of information an intelligent robot needs to adaptively and dexterously perform everyday activities. In our framework, every time a robot performs an activity, it creates an episodic memory. It can also acquire experiences from mental simulations, learn from these real and simulated experiences, and share them with other robots through dedicated knowledge web services.