From Human Hands to Robotic Limbs: A Study in Motor Skill Embodiment for Telemanipulation

 We proposed a novel teleoperation system for controlling a redundant degree-of-freedom (DOF) robot manipulator using human arm gestures.

Utilizing a GRU-based Variational Autoencoder (VAE) to learn a latent representation of the manipulator’s configuration space, capturing its complex joint kinematicsU. A fully-connected neural network maps human arm configurations into this latent space, allowing the system to mimic and generate corresponding manipulator trajectoris in real-time through the VAE decoder.