Feature-Space Reinforcement Learning for Robotic Manipulation

EasyChair Preprint 11420

Abstract

Reinforcement Learning (RL) has gained popular- ity for developing intelligent robots, but challenges such as sam- ple inefficiency and lack of generalization persist. The choice of observation space significantly influences RL algorithms’ sample efficiency in robotics. While end-to-end learning has been emphasized, it increases complexity and inefficiency as the agent must re-learn forward and inverse kinematics. To address these issues, we propose a straightforward approach that utilizes readily available control techniques, such as forward and inverse kinematics, to capitalize on domain knowledge. Our approach involves enhancing the observation space with task- space features and utilizing task-space inverse kinematics. Our contributions include a proposal for mathematical formulation and a framework for RL algorithms in robotics.

Keyphrases: Reinforcement Learning, feature space, robotics application

@booklet{EasyChair:11420,
  author    = {Ralf Gulde and Khanh Nguyen and Marc Tuscher and Oliver Riedel and Alexander Verl},
  title     = {Feature-Space Reinforcement Learning for Robotic Manipulation},
  howpublished = {EasyChair Preprint 11420},
  year      = {EasyChair, 2023}}