Human State Recognition Method in Intelligent Space

摘要

The Intelligent Space (iSpace) is a space that can analyze human behavior through computer vision and present user-centric information, and autonomously control robots as users intend. This study aims to support humans and robots to enhance their performances informatively and physically. In our previous research, we developed some robots and devices for iSpace, but they alone cannot fully support people. Cooperating with the robots and devices in the space with each other and deciding their optimal control considering the human state and surrounding conditions leads to better human assistance. To enable the cooperation, we propose a platform that can observe and store human state data. As human state, we recognize human gestures, facial expressions, and actions and integrate them as the human state. By mapping the human state to an environmental map, iSpace can learn information like user interest and utilization of space, and infer better robot control for users. We evaluated the recognition accuracy of each human state and validated the effect of our proposed method.