Master thesis: Learning Human-Robot Co-Manipulation from Human-Human Demonstrations via Vision-Based Diffusion Policy

Master thesis: Learning Human-Robot Co-Manipulation from Human-Human Demonstrations via Vision-Based Diffusion Policy

This thesis investigates learning-based human–robot co-manipulation for collaborative carrying tasks. A conditional Diffusion Transformer is developed to predict the robot follower’s future trajectory from 3D human pose observations and past robot motion. Human motion is captured using an RGB-D…
PyTorchDiffusion ModelsROS 2C++YOLO PoseMuJoCoFranka PandaRGB-DImpedance Control