Embodied AI - Master Thesis

You’ll work on our Embodied AI systems: real robots, in the real world. The work spans across the entire Embodied AI stack and includes learning, prototyping, integration, working with hardware and software. We frame the thesis around a concrete question we agree on together, but the bar is a working demo, not only a write-up. We particularly welcome EPFL students.

What you bring is eagerness, and expertise or experience in any discipline that might be useful. A few signals we like:

• You’ve built something that other people actually used.
• You’ve shown and demoed your own work in front of others.
• Hardware: CAD, 3D printing, PCB design, mechatronics integration (servos, sensors, cameras).
• Software: Linux, Docker, ROS2, Python, Git, React/TypeScript.
• Compute: Cloud (gcloud, AWS, Azure), Edge (RPi, Jetson), Microcontrollers (Arduino).
• Learning: IL/RL, PyTorch, Isaac ROS/Lab/Newton, W&B, HF’s LeRobot.