A Framework for Human Motion Estimation using IMUs in Human-Robot Interaction

Abstract

The field of human-robot interaction (HRI) covers a wide range of interactions between humans and robots in various settings. In most cases, both the human and the robot should know the status of each other and interchange data accordingly. Particularly, informing the robot about the human poses and gestures is critical for safe and reliable interaction. We present an open-source inertial measurement unit (IMU)-based human motion estimation framework designed for HRI applications which can be used with multiple robotic systems. The presented framework is developed as a Robot Operating System (ROS) package and takes on a bridge role between the human and the robot by utilizing the estimated human motions to produce a desired robot goal pose. Although there are different human motion tracking systems that are used in gaming, film making, industrial and medical applications today, they are either quite costly, tedious to set up or system dependent. Also, those systems have some limitations to be used in industrial environments such as suffering from occlusion and obstacles between the tracker system and the human, sensitive to light changes etc., and require long calibration steps before each use. The availability of a versatile human motion estimation framework that can be easily used in different HRI scenarios is handy for industrial applications. Some examples of usages of the proposed framework both in simulation and real-world applications are demonstrated in this paper. We aim our package to be a useful boosting tool to develop IMU-based human motion estimated HRI applications and auxiliary templates to develop more complex HRI scenarios for research and development purposes.