Skip to content

Latest commit

 

History

History
20 lines (15 loc) · 1.17 KB

SRL.md

File metadata and controls

20 lines (15 loc) · 1.17 KB

S-RL Toolbox: Environments, Datasets and Evaluation Metrics for State Representation Learning

Antonin Raffin [email protected] Ashley Hill [email protected] René Traoré [email protected] Timothée Lesort [email protected] Natalia Díaz-Rodríguez [email protected] David Filliat [email protected] U2IS, ENSTA ParisTech / INRIA FLOWERS Team http: // flowers. inria. fr

Abstract

State representation learning aims at learning compact representations from raw observations in robotics and control applications.

Approaches used for this objective are autoencoders, learning forward models, inverse dynamics or learning using generic priors on the state characteristics.

However, the diversity in applications and methods makes the field lack standard evaluation datasets, metrics and tasks.

This paper provides a set of environments, data generators, robotic control tasks, metrics and tools to facilitate iterative state representation learning and evaluation in reinforcement learning settings.

Keywords: Deep learning, reinforcement learning, state representation learning, robotic priors