Online motion generation for mirroring human arm motion

Roman Weitschat, Alexander Dietrich, Jorn Vogel
2016 2016 IEEE International Conference on Robotics and Automation (ICRA)  
Motion planning in robotics is a very large field of research. Many different approaches have been developed to create smooth trajectories for robot movement. For example there are optimization algorithms, which optimize kinematic or dynamic properties of a trajectory. Furthermore, nonlinear programming methods like e.g. optimal control, or polynomial based methods are widely used for trajectory generation. Most of these techniques are used to calculate a trajectory in advance, or they are
more » ... e, or they are limited to create point-to-point motions, where the robot needs to stop when switching to the next target point, especially, when interpolating in rotational space. In this paper, we combine a low-pass filter and spherical linear interpolation to realize a velocity-limited online trajectory generator for robot orientations in quaternion space. We use the developed motion generator for mirroring a human arm motion with a robot, recorded by a low frequency visual tracking. Using the proposed method, we can replicate the motion of the operator's arm with very little delay and thereby achieve an easy-to-use interface. Furthermore, as we can strictly limit the velocity of the generated motion, the approach can safely be used in human robot collaboration applications. * Joern Vogel and Roman Weitschat contributed equally to this work. All authors are with
doi:10.1109/icra.2016.7487620 dblp:conf/icra/WeitschatDV16 fatcat:bc4l4mof3ncsbm62zx26wd4q4y