arXiv Open Access 2026

Nightmare Dreamer: Dreaming About Unsafe States And Planning Ahead

Oluwatosin Oseni Shengjie Wang Jun Zhu Micah Corah

Lihat Sumber

Abstrak

Reinforcement Learning (RL) has shown remarkable success in real-world applications, particularly in robotics control. However, RL adoption remains limited due to insufficient safety guarantees. We introduce Nightmare Dreamer, a model-based Safe RL algorithm that addresses safety concerns by leveraging a learned world model to predict potential safety violations and plan actions accordingly. Nightmare Dreamer achieves nearly zero safety violations while maximizing rewards. Nightmare Dreamer outperforms model-free baselines on Safety Gymnasium tasks using only image observations, achieving nearly a 20x improvement in efficiency.

Topik & Kata Kunci

cs.LG cs.RO

Penulis (4)

Oluwatosin Oseni

Shengjie Wang

Jun Zhu

Micah Corah

Format Sitasi

APA MLA BibTeX

Oseni, O., Wang, S., Zhu, J., Corah, M. (2026). Nightmare Dreamer: Dreaming About Unsafe States And Planning Ahead. https://arxiv.org/abs/2601.04686

Akses Cepat

Lihat di Sumber

Informasi Jurnal

Tahun Terbit: 2026
Bahasa: en
Sumber Database: arXiv
Akses: Open Access ✓