Abstract In order to evaluate the engagement of interactive objects in human-robot interaction, an approach of engagement evaluation based on Dynamic Bayesian Network is proposed. First, the reasoning process of Dynamic Bayesian Network is described, and the topology of the evaluation model of the engagement is developed. Then, according to the limitation of parameterization of model based on data-driven, a parameterization process of model is designed by linguistic variables, fuzzification and defuzzification. The construction of evaluation model is completed. Finally, through the robot platform NAO and the design of interactive scenario, the evaluation model is verified. The experimental results show that the proposed method can make the robot recognize the interaction intention of the interactive object correctly in the process of human-robot interaction, and has high practical value and strong generalization ability.
Received: 19 June 2017
Published: 14 September 2017