Tactile-Aware Reinforcement Learning for Contact-Rich Manipulation

About the role

This thesis combines investigation and prototype development in tactile‑guided robotic manipulation. You will explore how tactile sensing can improve performance and robustness of contact‑rich tasks using a digital twin in NVIDIA Isaac Sim.

Key responsibilities

• Conduct a literature review on tactile sensing, dexterous manipulation and robot learning methods.
• Develop tactile sensing simulation based on contact information in Isaac Sim.
• Implement and train reinforcement‑learning policies for tasks such as grasp stabilization, pick‑and‑place and insertion.
• Compare tactile‑aware policies against proprioception‑only baselines and evaluate success rate, robustness and sim‑to‑real transfer.

Required profile

• Master student interested in robotics and reinforcement learning.
• Solid fundamentals in robotics (kinematics, dynamics, control) and basic machine learning.

Required skills

• Python programming.
• C++ programming.
• ROS 2 experience (preferred).
• Linux and Git proficiency.
• Familiarity with NVIDIA Isaac Sim / Isaac Lab.