Research on Interaction Migration Method of Service Dog-Human-Service Robot Based on MR Technology

The interaction models between humans and natural organisms have evolved over thousands of years. This study explores insights into robot control methods and the possibility of interaction method migrate from the natural interaction between humans and service dogs. Experimental data were obtained from interactions between humans and real service dogs, and a mixed reality (MR) experimental environment was established to collect control interaction data. Using Bayesian approach, the data were trained to construct an intent prediction model for human-service dog interaction. This model was then deployed in a human-quadruped robot control interaction system based on MR. Participants were recruited to conduct comparative experiments to verify the model's effectiveness. Results demonstrated that the human-robot control interaction system with the deployed prediction model outperformed systems with pure speech interaction system or multimodal control system without the model in terms of operational efficiency, user emotional experience, and usability. The intent prediction model developed in this study can be applied to various robot control scenarios, such as multi-robot control and emotional companion robot control, enhancing interaction efficiency and fostering emotional connections between humans and robots. Additionally, the MR first-person control system developed in this study can serve as a control hub for smart home devices and robots, providing more possibilities for the integration of multiple interaction methods.