Generating and Customizing Robotic Arm Trajectories using Neural Networks

Abstract

Common work of UKBA and IIT introduces a novel neural network-based method to create precise, smooth, and customizable robotic arm trajectories without requiring large datasets. Applied to the NICO humanoid robot (UHAM), the approach enables accurate pointing and gesture movements, paving the way for safer, more predictable, and human-friendly human-robot interaction. Explore the paper and code: GitHub Repository