In order for socially assistive robots to be successfully integrated within society, they need to be able to respond appropriately to the user during assistive human-robot interactions (HRI). Such robots are expected to interact with humans using natural interfaces. In this thesis, a novel robot emotion model that can be used for social robots engaged in HRI is presented. The proposed model effectively determines the robot’s emotional state based on its own emotion history, the affect of the user whom the robot is interacting with, and the HRI task at hand. The model uniquely uses an n^th order Markov Model (MM) to track the robot’s emotion history during interactions. Simulated experiments were conducted using the robot emotion model to persuade different users to comply with various tasks. A human-robot experiment with a social robot as an exercise coach was also designed to test the proposed model.