安全強化学習における安全な探索の再検討
Revisiting Safe Exploration in Safe Reinforcement learning
安全強化学習の安全性評価に「連続コストの期待最大値(EMCC)」という新指標を導入し、探索中の危険度をより適切に評価できるようにした。
著者: David Eckel, Baohe Zhang, Joschka Bödecker
分類: cs.LG, cs.AI, cs.RO
原文アブストラクト
Safe reinforcement learning (SafeRL) extends standard reinforcement learning with the idea of safety, where safety is typically defined through the constraint of the expected cost return of a trajectory being below a set limit. However, this metric fails to distinguish how costs accrue, treating infrequent severe cost events as equal to frequent mild ones, which can lead to riskier behaviors and result in unsafe exploration. We introduce a new metric, expected maximum consecutive cost steps (EMCC), which addresses safety during training by assessing the severity of unsafe steps based on their consecutive occurrence. This metric is particularly effective for distinguishing between prolonged and occasional safety violations. We apply EMMC in both on- and off-policy algorithm for benchmarking their safe exploration capability. Finally, we validate our metric through a set of benchmarks and propose a new lightweight benchmark task, which allows fast evaluation for algorithm design.
