Cloth folding with deep reinforcement learning

Speakers: Julius Hietala

Robotics Seminar Series. Tenth Session – 11th September 2020, 15:00-16:00, via zoom (use this link):


Robotic manipulation of soft objects such as cloth has not yet seen the same kind of progress that has been made in manipulating rigid objects in recent years. One of the main reasons for this is that soft objects have many degrees of freedom compared to rigid objects and thus they are harder to model and simulate. In this seminar, we will present reinforcement learning-based approaches for folding cloth in simulation and discuss future work related to transferring the learned policies to a real robot.

  • Jangir, Rishabh, Guillem Alenya, and Carme Torras. “Dynamic Cloth Manipulation with Deep Reinforcement Learning.” ArXiv Preprint ArXiv:1910.14475, 2019.
  • Matas, Jan, Stephen James, and Andrew J. Davison. “Sim-to-Real Reinforcement Learning for Deformable Object Manipulation.” ArXiv Preprint ArXiv:1806.07851, 2018.