TY  - RPRT
TI  - Compositional shield synthesis for safe reinforcement learning in partial observability
AU  - Steven Carr
AU  - Georgios Bakirtzis
AU  - Ufuk Topcu
PY  - 2025
UR  - https://arxiv.org/abs/2509.12085
ID  - 2509.12085
ER  -