A HUMAN-ROBOT INTERACTION CONTROL ARCHITECTURE FOR AN INTELLIGENT ASSISTIVE ROBOT

As the first round of baby boomers turn 65 in 2011, we must be prepared for the largest demographic group in history that could need long term care from nursing homes and home health providers. The development of socially assistive robots for health care applications can provide measurable improvements in patient safety, quality of care, and operational efficiencies by playing an increasingly important role in patient care in the fast pace of crowded clinics, hospitals and nursing/veterans homes. However, there are a number of research issues that need to be addressed in order to design such robots. In this paper, we address one of the main limitations to the development of intelligent socially assistive robots for health care applications: Robotic control architecture design and implementation with explicit social and assistive task functionalities. In particular, we present the design of a unique learning-based multi-layer decision making control architecture for utilization in determining the appropriate behavior of the robot. Herein, we explore and compare two different learning-based techniques that can be utilized as the main decision-making module of the controller. Preliminary experiments presented show the potential of the integration of the aforementioned techniques into the overall design of such robots intended for assistive scenarios.