Design Principles for Building Robust Human-Robot Interaction Machine Learning Models

Effective collaboration between humans and robots hinges on the robot's ability to comprehend its human teammate. This collaboration demands the development of machine learning models that bridge the gap between human physiological signals and their mental states. However, the challenge lies in developing generalizable machine learning models using data collected in controlled experimental conditions. This manuscript proposes a set of principles for designing human subject evaluations, emphasizing the crucial balance between experimental control and ecological validity while also balancing fundamental machine learning trade-offs.