Dataset-Driven Diffusion Policies for (Bi-)manipulator Imitation Learning

Completed
One-week Hackathon Project TUM Learning Systems and Robotics Lab x Poke and Wiggle
Robotics Imitation Learning Diffusion Models Manipulation

Collected human demonstrations for a LEGO Duplo stacking task using a VR-setup (Metaverse headset), learned Diffusion Transformer Flow policies based on more than 100 demonstrations and deployed the policies on a Franka Emika Panda.