Fault Tolerant Control

Fault-tolerant control (FTC) focuses on designing systems that can maintain functionality despite component failures, a critical need for robots and autonomous vehicles operating in unpredictable environments. Current research emphasizes developing robust control algorithms, often leveraging machine learning (e.g., reinforcement learning, genetic algorithms) and adaptive control strategies, to handle actuator failures (e.g., in robotic limbs or drone rotors) and sensor malfunctions. These advancements enhance the reliability and safety of robotic systems across diverse applications, from aerial vehicles and legged robots to underwater vehicles and industrial manipulators, improving their resilience and operational capabilities in challenging conditions.

Papers

April 17, 2022

Beta Residuals: Improving Fault-Tolerant Control for Sensory Faults via Bayesian Inference and Precision Learning
Mohamed Baioumy, William Hartemink, Riccardo M. G. Ferrari, Nick Hawes
Bayesian Inference Fault Detection Sensory Input Fault Tolerant Control Model Degradation Accurate Learning

November 26, 2021

Testability-Aware Low Power Controller Design with Evolutionary Learning
Min Li, Zhengyuan Shi, Zezhong Wang, Weiwei Zhang, Yu Huang, Qiang Xu
Fault Tolerant Control XAI System Evolutionary Machine Learning Based Testing

November 19, 2021

Reinforcement Learning with Adaptive Curriculum Dynamics Randomization for Fault-Tolerant Robot Control
Wataru Okamoto, Hiroshi Kera, Kazuhiko Kawamoto
Reinforcement Learning Robotic System Curriculum Reinforcement Learning Fault Tolerant Control Lloyd Algorithm Aware Curriculum Actuator Fault

Fault Tolerant Control

Papers

Beta Residuals: Improving Fault-Tolerant Control for Sensory Faults via Bayesian Inference and Precision Learning

Testability-Aware Low Power Controller Design with Evolutionary Learning

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