REFORMA is the first robust RL approach that uses adaptive adversaries to tackle uncertain disturbances in drone tasks.

In this work, we propose a robust RL approach via adaptive adversary (i.e., REFORMA) that allows the result- ing drone controller to generalize well toward ...

Aug 11, 2024 · REFORMA: Robust REinFORceMent Learning via Adaptive Adversary for Drones Flying under Disturbances ... learning via adversarial training ...

May 13, 2024 · "REFORMA: Robust REinFORceMent Learning via Adaptive Adversary for Drones Flying under Disturbances". Country unknown/Code not available: IEEE.

Luo, J. Dong, V. Tarokh, and Miroslav Pajic, “REFORMA: Robust REinFORceMent Learning via Adaptive Adversary for Drones Flying under Disturbances”, IEEE ...

Oct 11, 2023 · View a PDF of the paper titled Robust Safe Reinforcement Learning under Adversarial Disturbances, by Zeyang Li and 4 other authors ... adversary ( ...

Robust reinforcement learning through efficient adversarial ... REFORMA: Robust REinFORceMent Learning via Adaptive Adversary for Drones Flying under Disturbances.

Safe and robust RL: I study improving safety via unsupervised action planning and robustness via adversarial learning ( adversarial herding, adaptive adversary, ...

Aug 20, 2024 · REFORMA: Robust REinFORceMent Learning via Adaptive Adversary for Drones Flying under Disturbances. ... Learning through Efficient Adversarial ...

i.e., the size of adversary herd is upper-bounded to obtain sufficient approximation precision. optimal adversary. Adversarial Herd with Optimization Over Worst ...