This paper presents a human-agent interaction framework for emotional regulation through emotional interaction with a virtual agent that uses a consensus-based algorithm. This model-based framework provides a parameterized and dynamic approach for emotional interaction with an agent that can be applied to social robot-based therapies and virtual agent-based socio-emotional interaction studies. The extended goal of this study is to develop a human-robot interaction framework that can implement emotion regulation (ER) for children with Autism spectrum disorder (ASD). This paper describes the algorithmic approach for the design of this framework, as well as a user study to evaluate its effectiveness as a tool to provide emotional interaction and emotion regulation to children with ASD.