Whole-Body Teleoperation & Imitation Learning Thesis

About the role

ABB Robotics is seeking a master’s student to conduct a thesis on whole-body teleoperation and imitation learning for the Unitree G1 humanoid robot. The project combines AR‑based teleoperation, data collection, and policy training to enable autonomous robot behaviours.

Key responsibilities

• Review state‑of‑the‑art literature on whole‑body teleoperation, motion retargeting and imitation learning.
• Integrate an AR headset (e.g., Meta Quest) and leg‑worn IMUs into a real‑time teleoperation system that maps human full‑body pose to robot joints.
• Collect demonstration data of complex tasks (walking while reaching, pick‑and‑place) and store it for training.
• Train and validate imitation‑learning models in simulation (Isaac Sim) and on the real robot.

Required profile

• Enrolled in a Master’s program in Robotics, Computer Science, Electrical Engineering, Mechatronics or a related field.
• Strong programming skills in Python and/or C++.
• Basic knowledge of robot kinematics and control.
• Fluent in English (written and spoken).

Required skills

• Python
• C++
• PyTorch (or similar deep‑learning framework)
• Experience with physics simulators such as MuJoCo or Isaac Sim (preferred)
• Familiarity with AR headsets and inertial measurement units (IMU)