You will participate in testing, improving, and evaluating next-generation AI controllers that combine Imitation Learning and Reinforcement Learning, applied to real-robot arm manipulation.
Your responsibilities will include:
Collecting and annotating demonstrations via teleoperation.
Training AI controllers for Imitation Learning from these data.
Testing the algorithms on physical robots.
Creating and supervising a Reinforcement Learning setup in the real-world environment.
Contributing to the development of simulation, training, and evaluation tools.
Documenting your experiments and sharing insights with the technical team.
Note: The internship is not purely coding-focused; a substantial portion of manual teleoperation is part of the process to develop new robotic tasks (4 to 10 hours per new task).