Unmanned systems: a lab-based robotic arm for grasping phase II

Abstract

A 6 Degrees of Freedom (DOF) Leap Motion Controller (LMC) was characterized in position and accuracy for robotic arm control. Tests were conducted with linear and planar trajectories for input into the Kinova Jaco interface. The objective was to produce an intuitive and adaptive system that mimicked natural hand motion. Algorithms in C++ were produced to translate the LMC Cartesian position information to the Jaco Arm reference frame. Data showed that the LMC detector was quite sensitive to human hand jitter. Post-processing low-pass Fast Fourier Transform (FFT) filter techniques were employed to mitigate this problem. The LMC hand motion volume parameters were empirically scaled, in the Cartesian frame, to match Jaco motion operational requirements. It was determined that the LMC can be successfully used as an input device for the Jaco robotic arm control. Robotic arm trajectory latency issues were negligible when the Jaco Arm parameters for displacement trajectory rates were not violated and this was successfully managed in program code and user visual input.