Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Hashimoto, Kazumune, Kimura, Shunki, Serizawa, Kazunobu, Ikemoto, Junya, Gao, Yulong, Cai, Kai
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2604.02727
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917382659768320
author Hashimoto, Kazumune
Kimura, Shunki
Serizawa, Kazunobu
Ikemoto, Junya
Gao, Yulong
Cai, Kai
author_facet Hashimoto, Kazumune
Kimura, Shunki
Serizawa, Kazunobu
Ikemoto, Junya
Gao, Yulong
Cai, Kai
contents We study data-driven computation of probabilistic controlled invariant sets (PCIS) for safety-critical reinforcement learning under unknown dynamics. Assuming a linear MDP model, we use regularized least squares and self-normalized confidence bounds to construct a conservative estimate of the states from which the system can be kept inside a prescribed safe region over an \(N\)-step horizon, together with the corresponding set-valued safe action map. This construction is obtained through a backward recursion and can be interpreted as a conservative approximation of the \(N\)-step safety predecessor operator. When the associated conservative-inclusion event holds, a conservative fixed point of the approximate recursion can be certified as an \((N,ε)\)-PCIS with confidence at least \(η\). For continuous state spaces, we introduce a lattice abstraction and a Lipschitz-based discretization error bound to obtain a tractable approximation scheme. Finally, we use the resulting conservative fixed-point approximation as a runtime candidate PCIS in a practical shielding architecture with iterative updates, and illustrate the approach on a numerical experiment.
format Preprint
id arxiv_https___arxiv_org_abs_2604_02727
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Data-Driven Synthesis of Probabilistic Controlled Invariant Sets for Linear MDPs
Hashimoto, Kazumune
Kimura, Shunki
Serizawa, Kazunobu
Ikemoto, Junya
Gao, Yulong
Cai, Kai
Systems and Control
We study data-driven computation of probabilistic controlled invariant sets (PCIS) for safety-critical reinforcement learning under unknown dynamics. Assuming a linear MDP model, we use regularized least squares and self-normalized confidence bounds to construct a conservative estimate of the states from which the system can be kept inside a prescribed safe region over an \(N\)-step horizon, together with the corresponding set-valued safe action map. This construction is obtained through a backward recursion and can be interpreted as a conservative approximation of the \(N\)-step safety predecessor operator. When the associated conservative-inclusion event holds, a conservative fixed point of the approximate recursion can be certified as an \((N,ε)\)-PCIS with confidence at least \(η\). For continuous state spaces, we introduce a lattice abstraction and a Lipschitz-based discretization error bound to obtain a tractable approximation scheme. Finally, we use the resulting conservative fixed-point approximation as a runtime candidate PCIS in a practical shielding architecture with iterative updates, and illustrate the approach on a numerical experiment.
title Data-Driven Synthesis of Probabilistic Controlled Invariant Sets for Linear MDPs
topic Systems and Control
url https://arxiv.org/abs/2604.02727