📄 中文摘要
传统的人工智能对齐主要关注单个模型输出，而在长时程工作流中，自主代理需要在整个交互轨迹中保持持续的可靠性。提出了一种名为APEMO（情感感知峰值-结束调制的调度层）的运行时调度层，旨在通过操作时间-情感信号来优化固定预算下的计算分配。APEMO并不修改模型权重，而是通过行为代理检测轨迹不稳定性，并针对关键片段（如峰值时刻和结束时刻）进行修复。通过多代理模拟和基于大型语言模型的规划-执行流的评估，APEMO在轨迹级质量和重用概率方面表现出一致的提升。
时间对齐：面向长时程自主系统的峰值感知调度
📄 English Summary
Alignment in Time: Peak-Aware Orchestration for Long-Horizon Agentic Systems
The study presents APEMO (Affect-aware Peak-End Modulation for Orchestration), a runtime scheduling layer designed to optimize computational allocation under fixed budgets by operationalizing temporal-affective signals. Traditional AI alignment primarily focuses on individual model outputs; however, autonomous agents in long-horizon workflows require sustained reliability across entire interaction trajectories. APEMO detects trajectory instability through behavioral proxies and targets repairs at critical segments, such as peak moments and endings, rather than modifying model weights. Evaluation across multi-agent simulations and LLM-based planner-executor flows demonstrates that APEMO consistently enhances trajectory-level quality and reuse probability.