/Hindsight-Experience-Replay

Hindsight Experience Replay algorithm research

Primary LanguagePython

Hindsight Experience Replay

Controlling a Spaceship using Hindsight Experience Replay (a.k.a HER)

This research is based on the paper Hindsight Experience Replay submitted on Jul 5th, 2017 by OpenAI Researchers.

I wrote a series of Medium articles trying to demystify this algorithm, where I describe my journey during the reaserch.

I'm using Deep Q-Network with Double DQN and Dueling Network Architecture.

Abstract

Dealing with sparse rewards is one of the biggest challenges in Reinforcement Learning (RL). We present a novel technique called Hindsight Experience Replay which allows sample-efficient learning from rewards which are sparse and binary and therefore avoid the need for complicated reward engineering. It can be combined with an arbitrary off-policy RL algorithm and may be seen as a form of implicit curriculum.

We demonstrate our approach on the task of manipulating objects with a robotic arm. In particular, we run experiments on three different tasks: pushing, sliding, and pick-and-place, in each case using only binary rewards indicating whether or not the task is completed. Our ablation studies show that Hindsight Experience Replay is a crucial ingredient which makes training possible in these challenging environments. We show that our policies trained on a physics simulation can be deployed on a physical robot and successfully complete the task.

Resources

Papers

  1. Hindsight Experience Replay
  2. DHER: Hindsight Experience Replay for Dynamic Goals
  3. Hindsight policy gradients
  4. Rewriting History with Inverse RL: Hindsight Inference for Policy Improvement
  5. Advances in Experience Replay
  6. Curriculum-guided Hindsight Experience Replay
  7. Soft Hindsight Experience Replay

Articles

  1. Reinforcement Learning with Hindsight Experience Replay
  2. Learning from mistakes with Hindsight Experience Replay
  3. Advanced Exploration: Hindsight Experience Replay
  4. Understanding DQN+HER

Videos

  1. Hindsight Experience Replay | Two Minute Papers #192
  2. Overcoming sparse rewards in Deep RL: Curiosity, hindsight & auxiliary tasks
  3. Hindsight Experience Replay
  4. Hindsight Experience Replay by Olivier Sigaud

Repos

  1. 2D Gridworld navigation using RL with Hindsight Experience Replay
  2. Pytorch implementation of Hindsight Experience Replay (HER)
  3. Reproducing results from the Hindsight Experience Replay paper in PyTorch
  4. Hindsight Experience Replay by Alex Hermansson