Average-Reward Maximum Entropy Reinforcement Learning for Global Policy in Double Pendulum Tasks

2025/05/12 by Jean Seong Bjorn Choe, Choe, Jean Seong Bjorn, Bumkyu Choi +3
Computer Science · Engineering · #Adaptive Dynamic Programming Control #FOS: Computer and information sciences #Reinforcement Learning in Robotics #Robot Manipulation and Learning #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.2505.07516

openalex publication_date 2025/05/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This report presents our reinforcement learning-based approach for the swing-up and stabilisation tasks of the acrobot and pendubot, tailored specifcially to the updated guidelines of the 3rd AI Olympics at ICRA 2025. Building upon our previously developed Average-Reward Entropy Advantage Policy Optimization (AR-EAPO) algorithm, we refined our solution to effectively address the new competition scenarios and evaluation metrics. Extensive simulations validate that our controller robustly manages these revised tasks, demonstrating adaptability and effectiveness within the updated framework.

Citations

• Reinforcement Learning for Robust Athletic Intelligence: Lessons from the 2nd 'AI Olympics with RealAIGym' Competition
• Velocity-History-Based Soft Actor-Critic Tackling IROS'24 Competition "AI Olympics with RealAIGym"
• Average-Reward Maximum Entropy Reinforcement Learning for Underactuated Double Pendulum Tasks
• Learning control of underactuated double pendulum with Model-Based Reinforcement Learning
• AI Olympics challenge with Evolutionary Soft Actor Critic
• Proximal Policy Optimization Algorithms
• High-Dimensional Continuous Control Using Generalized Advantage\n Estimation

Related

• Average-Reward Maximum Entropy Reinforcement Learning for Underactuated Double Pendulum Tasks
• Learning control of underactuated double pendulum with Model-Based Reinforcement Learning
• PPO-BR: Dual-Signal Entropy-Reward Adaptation for Trust Region Policy Optimization
• Solving the swing-up and balance task for the Acrobot and Pendubot with SAC
• Reinforcement Learning for Robust Athletic Intelligence: Lessons from the 2nd 'AI Olympics with RealAIGym' Competition