A ROS-Integrated Skeleton Detection Framework Enabling Human-Centered Robotic Applications

The escalating demand for assistive robots, fueled by an aging population and the quest for technological solutions to enhance quality of life, requires an effective comprehension of data coming from users interacting with robots. This paper proposes a holistic solution to the challenge of human skeleton detection and analysis by integrating a 3D skeleton detector and RGBD camera spatial data within the Robot Operating System (ROS) middleware. This integration facilitates the publication of user pose and location, forming a versatile foundation for diverse assistive applications. Additionally, a whole-body joint analysis method based on analytical geometry is introduced, expanding the robot's understanding of the user's body and enabling a broader range of applications. The paper showcases the framework's versatility through two applications: humanoid robot pose retargeting and pose classification for user behavior study, proving that the proposed solution works in real time and serves as a modular, adaptable solution for diverse assistive robotic applications.