AI for Industry Challenge — cable insertion
Research internship · IL/RL policies and a sim-to-real stack for the Intrinsic & Open Robotics challenge
Robotics Research Intern, ETF Robotics Laboratory, University of Belgrade (Nov 2025 – Jul 2026). Challenge · Code
Our lab’s entry in the AI for Industry Challenge organized by Intrinsic and Open Robotics: teach a robot to insert cables into industrial connectors.
- Developed imitation- and reinforcement-learning policies for cable insertion in NVIDIA Isaac Lab.
- Built the simulation scenes in OpenUSD.
- Set up a containerized ROS 2 (Kilted Kaiju) / Gazebo environment to carry policies from simulation to the real robot.
Related: our ETRAN 2026 paper on bridging simulation and reality for robot assembly tasks (see publications).