Intelligent Robotics

Supervisor: Prof. Ville Kyrki (ville.kyrki(at)aalto.fi)

Advisors: MSc David Blanco Mulero (david.blancomulero(at)aalto.fi), Dr. Kevin Luck (kevin.s.luck(at)aalto.fi)

Keywords: adversarial learning, robotic manipulation, deformable object manipulation.

Project Description

Manipulation of deformable objects such as making a bed or folding a t-shirt are tasks still far from being solved. Recent approaches have tried to solve tasks like folding or unfolding a cloth via Reinforcement Learning or Supervised Learning. However, this methods are far from optimal for generalising to new shapes, sizes and materials. A tentative solution is to use a adversarial learning strategies to improve the robustness of these systems.

This thesis will focus on how to use adversarial learning for improving existing methods in robotic cloth manipulation. The thesis will explore the state-of-the-art and work on solutions for cloth manipulation.

Previous knowledge on Machine Learning is desired but not mandatory.

Deliverables

• Review relevant adversarial learning literature,
• Implementation of relevant adversarial and manipulation algorithms,
• Evaluation of the experimental results,
• Evaluation of algorithms on real robots.

Practical Information

Pre-requisites: Python (high/medium), Machine Learning (medium)

Tools: OpenAI gym, Softgym

Start: Available Immediately

References

Tsurumine, Yoshihisa, et al. “Generative adversarial imitation learning with deep p-network for robotic cloth manipulation.” 2019 IEEE-RAS 19th International Conference on Humanoid Robots (Humanoids). IEEE, 2019.  

Wang, David, et al. “Adversarial grasp objects.” 2019 IEEE 15th International Conference on Automation Science and Engineering (CASE). IEEE, 2019.

Morrison, Douglas, Peter Corke, and Jürgen Leitner. “Egad! an evolved grasping analysis dataset for diversity and reproducibility in robotic manipulation.” IEEE Robotics and Automation Letters 5.3 (2020): 4368-4375.

Ren, Allen Z., and Anirudha Majumdar. “Distributionally robust policy learning via adversarial environment generation.” IEEE Robotics and Automation Letters 7.2 (2022): 1379-1386.

Jiang, Junnan, et al. “Learning Grasp Ability Enhancement Through Deep Shape Generation.” International Conference on Intelligent Robotics and Applications. Springer, Cham, 2022.