Fault-Aware Robust Control via Adversarial Reinforcement Learning

2020/11/17 by Fan Yang, Chao Yang, Yang, Fan +7 · 1 citation
Computer Science · Engineering · #Adversarial Robustness in Machine Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Reinforcement Learning in Robotics #Robot Manipulation and Learning #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.2011.08728

openalex publication_date 2020/11/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Robots have limited adaptation ability compared to humans and animals in the case of damage. However, robot damages are prevalent in real-world applications, especially for robots deployed in extreme environments. The fragility of robots greatly limits their widespread application. We propose an adversarial reinforcement learning framework, which significantly increases robot robustness over joint damage cases in both manipulation tasks and locomotion tasks. The agent is trained iteratively under the joint damage cases where it has poor performance. We validate our algorithm on a three-fingered robot hand and a quadruped robot. Our algorithm can be trained only in simulation and directly deployed on a real robot without any fine-tuning. It also demonstrates exceeding success rates over arbitrary joint damage cases.

Citations

• Continuous control with deep reinforcement learning
• Robust Adversarial Reinforcement Learning
• Robust Deep Reinforcement Learning with Adversarial Attacks
• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
• Action Robust Reinforcement Learning and Applications in Continuous Control
• Learning to Adapt in Dynamic, Real-World Environments Through Meta-Reinforcement Learning

Cited by

• Reinforcement Learning with Adaptive Curriculum Dynamics Randomization for Fault-Tolerant Robot Control

Related

• Adaptive Compensation for Robotic Joint Failures Using Partially Observable Reinforcement Learning
• Reinforcement Learning with Adaptive Curriculum Dynamics Randomization for Fault-Tolerant Robot Control
• Deep Reinforcement Learning for Single-Shot Diagnosis and Adaptation in\n Damaged Robots
• Adversarial Fine-tuning in Offline-to-Online Reinforcement Learning for Robust Robot Control
• Robustness Evaluation of Offline Reinforcement Learning for Robot Control Against Action Perturbations