规划安全论文趋势 2023–2026

主要结论

• 规划安全论文从 2023 年的 33 篇增长到 2026 年的 37 篇，三年间扩大到约 1.1 倍。
• NeurIPS 是发规划安全论文最多的会议，2023–2026 年共 59 篇，占四个会议总量的 56%。
• 交叉方向里涨得最快的是社会层面 · safety：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 5 篇。
• 在「博弈与规划」的 5 个主题里，它的 2026 年论文数排第 5；同组增长最快的经典规划是 3.3 倍，规划安全是 1.1 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026PASSAGE: A Real-Terrain Benchmark for Trustworthy Constrained Path Planning
• ICML 2026Constrained Bayesian Experimental Design via Online Planning
• ICLR 2026Superficial Safety Alignment Hypothesis
• CVPR 2026ActiveAD Planning-Oriented Active Learning for End-to-End Autonomous Driving
• NeurIPS 2025Refusal Direction is Universal Across Safety-Aligned Languages
• ICML 2025Goal-Space Planning with Subgoal Models
• ICLR 2025Durable Quantization Conditioned Misalignment Attack on Large Language Models
• NeurIPS 2024Introspective Planning: Aligning Robots' Uncertainty with Inherent Task Ambiguity
• ICML 2024MC-GTA: Metric-Constrained Model-Based Clustering using Goodness-of-fit Tests with Autocorrelations
• ICLR 2024Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions
• CVPR 2024Anatomically Constrained Implicit Face Models
• NeurIPS 2023Experiment Planning with Function Approximation
• ICML 2023Constrained Monotonic Neural Networks
• ICLR 2023Integrating Symmetry into Differentiable Planning with Steerable Convolutions
• NeurIPS 2026Ordinal Geometry Complements Reconstruction: Diagnosing Planning with Compressed Value Functions
• ICML 2026On the Ability of Transformers to Verify Plans
• ICLR 2026Mitigating the Safety Alignment Tax with Null-Space Constrained Policy Optimization
• CVPR 2026HQC-NBV A Hybrid Quantum-Classical View Planning Approach
• NeurIPS 2025Alignment of Large Language Models with Constrained Learning
• ICML 2025Training Deep Learning Models with Norm-Constrained LMOs

全部规划安全论文

共 14 篇，第 1 / 1 页

1. ICLR 2026Mitigating the Safety Alignment Tax with Null-Space Constrained Policy Optimization
2. ICLR 2026SAFETY-GUIDED FLOW (SGF): A UNIFIED FRAMEWORK FOR NEGATIVE GUIDANCE IN SAFE GENERATION
3. ICLR 2026Superficial Safety Alignment Hypothesis
4. ICLR 2026When a Robot is More Capable than a Human: Learning from Constrained Demonstrators
5. ICLR 2025ActSafe: Active Exploration with Safety Constraints for Reinforcement Learning
6. ICLR 2025Conservative Contextual Bandits: Beyond Linear Representations
7. ICLR 2025Durable Quantization Conditioned Misalignment Attack on Large Language Models
8. ICLR 2025SafeDiffuser: Safe Planning with Diffusion Probabilistic Models
9. ICLR 2025Safety-Prioritizing Curricula for Constrained Reinforcement Learning
10. ICLR 2024GRAPH-CONSTRAINED DIFFUSION FOR END-TO-END PATH PLANNING
11. ICLR 2024Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions
12. ICLR 2023Integrating Symmetry into Differentiable Planning with Steerable Convolutions
13. ICLR 2023Learning Diffusion Bridges on Constrained Domains
14. ICLR 2023Safe Reinforcement Learning From Pixels Using a Stochastic Latent Representation

常见问题

2023–2026 年四大会议一共有多少篇规划安全论文？

共 105 篇，按年份分别是 2023 年 33 篇、2024 年 16 篇、2025 年 19 篇、2026 年 37 篇。

NeurIPS 2026 有多少篇规划安全论文？

NeurIPS 2026 有 21 篇规划安全论文，上一届是 7 篇。

ICLR 2026 有多少篇规划安全论文？

ICLR 2026 有 4 篇规划安全论文，上一届是 5 篇。

ICML 2026 有多少篇规划安全论文？

ICML 2026 有 8 篇规划安全论文，上一届是 7 篇。

CVPR 2026 有多少篇规划安全论文？

CVPR 2026 有 4 篇规划安全论文。

哪个会议的规划安全论文最多？

NeurIPS，2023–2026 年共 59 篇。