In this paper, we present an approach for online teleoperation of anthropomorphic robotic arms via hand and arm movement. Our setup enables 7DoF arm tracking and hand gesture recognition at high frequencies. The approach provides a simple and intuitive way of controlling all joints of a robotic arm and additionally offers the possibility to control robot end-effectors through hand gestures. The setup is inexpensive and easy to put up. Our installation is composed of multiple depth cameras that are matched to one another through a 3D calibration procedure. Thus, higher point densities can be flexibly reached for different parts of the body by positioning the cameras. We demonstrate the feasibility of our approach by implementing and testing the setup on a KUKA LBR iiwa 7. We analyze the system's ergonomics by adapting the ISO 9241-9 standard and compare it to a traditional 3D mouse-based human-robot interface.
展开▼
