Investigation of dynamic 3D hand motion reproduction by a robot using a Leap Motion

Abstract

To increase their competitiveness, companies use more and more robotic systems to perform various complex tasks. These robots are very attractive because they are flexible and low-cost. A main current issue is the co-activity with human. With the very recent introduction of the new low-cost motion capture system Leap Motion, which has characteristics of accuracy unmatched to this cost for the moment, authors propose a first work concerning the real-time reproduction of 3D human motion by a robotic system. The aim is to reproduce complex tasks in 3D without any constraints for the operator. This device allows to generate a very accurate point cloud of positions and directions. To ensure a good repeatability of the movement on an UR10 robot, interpolation of these data by polynomial splines minimizing the L1 norm are done. This formalism has been developed recently and presents a linear complexity calculation with data and maintains the shape of the data even when the discretization is not uniform in space.