TY  - RPRT
TI  - Safe Continual Reinforcement Learning under Nonstationarity via Adaptive Safety Constraints
AU  - Timofey Tomashevskiy
PY  - 2026
UR  - https://arxiv.org/abs/2605.18842
ID  - 2605.18842
ER  -