Benchmarking Actor-Critic Deep Reinforcement Learning Algorithms for Robotics Control with Action Constraints

Kazumi Kasaura, Shuwa Miura, Tadashi Kozuno, Ryo Yonetani, Kenta Hoshino +1 more
4/18/2023
cs.LGcs.RO

No official code found. Best implementation: omron-sinicx/action-constrained-RL-benchmark (28 stars, MIT, last commit 2024-04-26).

Abstract

This study presents a benchmark for evaluating action-constrained reinforcement learning (RL) algorithms. In action-constrained RL, each action taken by the learning system must comply with certain constraints. These constraints are crucial for ensuring the feasibility and safety of actions in real-world systems. We evaluate existing algorithms and their novel variants across multiple robotics control environments, encompassing multiple action constraint types. Our evaluation provides the first in-depth perspective of the field, revealing surprising insights, including the effectiveness of a straightforward baseline approach. The benchmark problems and associated code utilized in our experiments are made available online at github.com/omron-sinicx/action-constrained-RL-benchmark for further research and development.

View on arXivView PDF

Code Implementations(1)

omron-sinicx/action-constrained-RL-benchmarkCommunity100%
MIT
284Shell, PythonApr 10, 20232 years agoMIT

Similar Papers

Taxonomy and Trends in Reinforcement Learning for Robotics and Control Systems: A Structured Review

87% match2025

Leveraging Constraint Violation Signals For Action-Constrained Reinforcement Learning

86% match1 repo12025

Efficient Action-Constrained Reinforcement Learning via Acceptance-Rejection Method and Augmented MDPs

86% match2025

Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes

86% match2024

CaT: Constraints as Terminations for Legged Locomotion Reinforcement Learning

86% match2024

Learning to Generate All Feasible Actions

85% match2023

Cite this paper

@article{kasaura2023benchmarking,
  title  = {Benchmarking Actor-Critic Deep Reinforcement Learning Algorithms for Robotics Control with Action Constraints},
  author = {Kazumi Kasaura and Shuwa Miura and Tadashi Kozuno and Ryo Yonetani and Kenta Hoshino and Yohei Hosoe},
  year   = {2023},
  eprint = {2304.08743},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2304.08743v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2304.08743.svg)](https://codeofpaper.com/paper/2304.08743)

Discussion