🛠 Tech Stack

💼 About This Role

You'll develop a learning-based controller for dual-arm robot manipulation in real-world warehouse conditions. You will apply imitation learning or diffusion policies to enable robots to grasp and manipulate parcels. This thesis offers hands-on experience with state-of-the-art robot learning and sim-to-real transfer.

🎯 What You'll Do

• Teleoperate two robot arms to collect demonstration data
• Train a policy using imitation learning or diffusion policies
• Transfer policy from simulation to real robot
• Evaluate robustness across parcel shapes and clutter

📋 Requirements

• Solid Python programming skills
• Familiarity with machine learning frameworks such as PyTorch
• Basic knowledge of robotics or computer vision
• Strong problem-solving skills

✨ Nice to Have

• Experience with graph neural networks
• Experience with 3D perception or point clouds
• Experience with physics simulation

📨 Hiring Process

Only open to current ETH Zurich CS Master students; apply for more details.