Multi Goal Reinforcement Learning

Multi-goal reinforcement learning (MRL) focuses on training agents to achieve multiple, potentially diverse goals within a single environment, improving efficiency and generalization compared to single-goal approaches. Current research emphasizes addressing challenges like the under-exploration of difficult goals, often employing techniques such as adaptive sampling, curriculum learning, and hindsight experience replay to improve sample efficiency and success rates. These advancements leverage various model architectures, including goal-conditioned Q-functions, generative adversarial networks (GANs), and bilinear value networks, to better represent and generalize across the goal space. MRL's impact spans robotics, where it enables more versatile and adaptable robots, and broader AI, offering insights into efficient learning and generalization in complex environments.

Papers

October 16, 2024

SAC-GLAM: Improving Online RL for LLM agents with Soft Actor-Critic and Hindsight Relabeling
Loris Gaven, Clement Romac, Thomas Carta, Sylvain Lamprier, Olivier Sigaud, Pierre-Yves Oudeyer
LLM Agent Online Reinforcement Learning Autonomous LLM Multi Goal Reinforcement Learning Hindsight Goal Relabeling

December 21, 2023

Open-Source Reinforcement Learning Environments Implemented in MuJoCo with Franka Manipulator
Zichun Xu, Yuntao Li, Xiaohang Yang, Zhiyuan Zhao, Lei Zhuang, Jingdong Zhao
Serial Manipulator Multi Agent MuJoCo Multi Goal Reinforcement Learning Open Source Reinforcement Learning

May 23, 2023

L-SA: Learning Under-Explored Targets in Multi-Target Reinforcement Learning
Kibeom Kim, Hyundo Lee, Min Whoo Lee, Moonheon Lee, Minsu Lee, Byoung-Tak Zhang
Reinforcement Learning Adaptive Sampling Multi Goal Reinforcement Learning

May 5, 2023

Composite Motion Learning with Task Control
Pei Xu, Xiumin Shang, Victor Zordan, Ioannis Karamouzas
Motion Model Motion Imitation Diverse Motion Multi Goal Reinforcement Learning Task Transition

September 26, 2022

Understanding Hindsight Goal Relabeling from a Divergence Minimization Perspective
Lunjun Zhang, Bradly C. Stadie
Imitation Learning Suboptimal Demonstration Hindsight Instruction Multi Goal Reinforcement Learning Divergence Optimization Hindsight Goal Relabeling

August 31, 2022

Cluster-based Sampling in Hindsight Experience Replay for Robotic Tasks (Student Abstract)
Taeyoung Kim, Dongsoo Har
Sparse Reward Robotic Task Experience Replay Sampling Strategy Hindsight Experience Replay Multi Goal Reinforcement Learning Based Sampling

July 22, 2022

Graph-Structured Policy Learning for Multi-Goal Manipulation Tasks
David Klee, Ondrej Biza, Robert Platt
Q Learning Robotic Manipulation Manipulation Task State Representation Multi Goal Reinforcement Learning

June 14, 2022

Stein Variational Goal Generation for adaptive Exploration in Multi-Goal Reinforcement Learning
Nicolas Castanet, Sylvain Lamprier, Olivier Sigaud
Stein Variational Gradient Descent Variational Objective Exploration Problem Adaptive Exploration Multi Goal Reinforcement Learning

June 9, 2022

Pragmatically Learning from Pedagogical Demonstrations in Multi-Goal Environments
Hugo Caselles-Dupré, Olivier Sigaud, Mohamed Chetouani
Video Demonstration Goal Conditioned Reinforcement Learning Pragmatic Reasoning Multi Goal Multi Goal Reinforcement Learning Goal Inference Optimal Demonstration

April 28, 2022

Bilinear value networks
Zhang-Wei Hong, Ge Yang, Pulkit Agrawal
Q Function Q Value Bilinear Matrix Inequality Multi Goal Reinforcement Learning Value Network

April 8, 2022

Grounding Hindsight Instructions in Multi-Goal Reinforcement Learning for Robotics
Frank Röder, Manfred Eppe, Stefan Wermter
Robotics Domain Hindsight Instruction Robot Policy Learning Goal Representation Multi Goal Reinforcement Learning

November 7, 2021

Automatic Goal Generation using Dynamical Distance Learning
Bharat Prakash, Nicholas Waytowich, Tinoosh Mohsenin, Tim Oates
Reinforcement Learning Markov Decision Process Dynamic Learning Multi Goal Reinforcement Learning Goal Generation

Multi Goal Reinforcement Learning

Papers

SAC-GLAM: Improving Online RL for LLM agents with Soft Actor-Critic and Hindsight Relabeling

Open-Source Reinforcement Learning Environments Implemented in MuJoCo with Franka Manipulator

L-SA: Learning Under-Explored Targets in Multi-Target Reinforcement Learning

Composite Motion Learning with Task Control

Understanding Hindsight Goal Relabeling from a Divergence Minimization Perspective

Cluster-based Sampling in Hindsight Experience Replay for Robotic Tasks (Student Abstract)

Graph-Structured Policy Learning for Multi-Goal Manipulation Tasks

Stein Variational Goal Generation for adaptive Exploration in Multi-Goal Reinforcement Learning

Pragmatically Learning from Pedagogical Demonstrations in Multi-Goal Environments

Bilinear value networks

Grounding Hindsight Instructions in Multi-Goal Reinforcement Learning for Robotics

Automatic Goal Generation using Dynamical Distance Learning