Dynamic, Predictive-Memory Architecture for Intelligent Agents

This project is aimed at building robotic assistants to work with humans in space environments. The key technical innovation is the software integration of a high-level memory architecture capable of deliberative planning and natural language understanding with a reactive execution system capable of real-time autonomous activity. Project effort will be in the design and implementation of the Dynamic Predictive Memory Architecture for an autonomous robot. The primary focus of phase I will be constructing a demonstration system in simulation and the study of NASA robotic systems and task domains. The goal is to consolidate existing research results on DPMA into a coherent development system and identify and design a working prototype system that can be implemented on real-hardware in phase II. We expect phase I and phase II to pave the way for the development of robotic assistants for humans engaged in a variety of real world tasks.