Low Dimensional State Representation Learning with Robotics Priors in Continuous Action Spaces

Reinforcement learning algorithms have proven to be capable of solving complicated robotics tasks in an end-to-end fashion without any need for hand-crafted features or policies. Especially in the context of robotics, in which the cost of real-world data is usually extremely high, Reinforcement Lear...

Full description

Saved in:
Bibliographic Details
Published in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) pp. 190 - 197
Main Authors Botteghi, Nicolo, Alaa, Khaled, Poel, Mannes, Sirmacek, Beril, Brune, Christoph, Mersha, Abeje, Stramigioli, Stefano
Format Conference Proceeding
LanguageEnglish
Published IEEE 27.09.2021
SeriesIEEE International Conference on Intelligent Robots and Systems
Subjects
Action spaces
Continuous actions
End to end
High-dimensional
Learning algorithms
Low dimensional
Mathematical models
Mobile robots
Navigation
Real-world
Reinforcement learning
Reinforcement learning algorithms
Reinforcement learning solution
Representation learning
Robot sensing systems
Robotic tasks
Robotics
State representation
Virtual environments
Virtual reality
Visualization
Online AccessGet full text

Cover

More Information
Summary:Reinforcement learning algorithms have proven to be capable of solving complicated robotics tasks in an end-to-end fashion without any need for hand-crafted features or policies. Especially in the context of robotics, in which the cost of real-world data is usually extremely high, Reinforcement Learning solutions achieving high sample efficiency are needed. In this paper, we propose a framework combining the learning of a low-dimensional state representation, from high-dimensional observations coming from the robot's raw sensory readings, with the learning of the optimal policy, given the learned state representation. We evaluate our framework in the context of mobile robot navigation in the case of continuous state and action spaces. Moreover, we study the problem of transferring what learned in the simulated virtual environment to the real robot without further retraining using real-world data in the presence of visual and depth distractors, such as lighting changes and moving obstacles. A video of our experiments can be found at: https://youtu.be/rUdGPKr2Wuo.
ISBN:1665417153
9781665417150
ISSN:2153-0866
DOI:10.1109/IROS51168.2021.9635936