Date of Award
August 2019
Degree Type
Thesis
Degree Name
Master of Science (MS)
Department
Electrical Engineering and Computer Science
Advisor(s)
Garrett E. Katz
Keywords
Artificial Intelligence, empathy, Markov Decision Process
Subject Categories
Engineering
Abstract
Among the different emotional functions, empathy is difficult to model but is important for a robot to have so that they can socially interact with society. Perceived empathy has positive consequences on attitude and the social behaviors of an individual. The project aims to model empathic behavior and perceptive capabilities in a robot in such a way that they can engage in empathic interactions with other agents in a shared physical space. In this study, a small grid environment world was designed and developed with multiple agents. Also, The Markov Decision Process and a Hand Coded algorithm were implemented and compared to study the influence of empathy on the robot agent’s behavior. In this environment when empathy, needy and food parameters are varied from -1 to 1, -1 to 1, and unlimited to limited supply, respectively, change is observed in the behavior of the robot agent. When the food supply is unlimited, the robot agent can either aid or avoid other agents based on the positive, zero or negative values of empathy and needy. In contrast, when food supply is limited, the robot agent’s behavior towards other agents varies based upon the greater of the two values.
Access
Open Access
Recommended Citation
Patel, Dhwani Himanshu, "EMPATHY BASED REINFORCEMENT LEARNING" (2019). Theses - ALL. 352.
https://surface.syr.edu/thesis/352
