Thesis: Active Vision for AI-based Robot Manipulation

About the role

This thesis investigates the benefits of a 2‑DOF movable camera head for AI‑based robot manipulation using Vision‑Language‑Action models. You will work through the full pipeline from data collection to model deployment, supported by the ABB team.

Key responsibilities

• Extend data collection software to support a movable head setup.
• Upgrade the teleoperation system with neck control using an AR headset.
• Collect and record robot manipulation demonstrations with the movable head.
• Train and evaluate VLA or similar models that incorporate head motion.
• Deploy the model on a real robot, handling synchronization and control delays.
• Compare performance with and without the movable camera.

Required profile

• Master's student in Computer Science, Electrical Engineering, Robotics, Mechatronics or related field.
• Strong interest in robot learning, machine learning, computer vision or physical AI.
• Ability to work on‑site in Västerås from January to July 2027.

Required skills

• Programming in Python.
• Experience with deep learning frameworks such as PyTorch and HuggingFace.
• Familiarity with LeRobot platform.
• Knowledge of git and Docker.
• Experience with Unity or AR tools is a plus.