CausalNav：用于物理参数偏移下控制的可靠性可验证因果世界模型
CausalNav: Reliability-Certified Causal World Models for Control under Physical-Parameter Shift
- University of Wollongong（卧龙岗大学）
- CSIRO’s Data61（联邦科学与工业研究组织Data61）
- Adelaide University（阿德莱德大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文提出CausalNav控制器，其结合因果转移图与多门控可靠性验证机制，在物理参数偏移的CartPole-v1和Pendulum-v1任务上，较9种基准取得最佳平均排名，关键在于经认证的弃权（不执行）提升了部署安全性。
AI中文摘要：
世界模型仅在智能体决策时能改变其行为才对物理AI有用，且仅在自身判断出错时会弃权（不执行）才具备安全性。本文围绕CausalNav展开研究，这是一种基于经识别状态坐标上带符号、动作条件化转移图的控制器。部署时，CausalNav会模拟少量干预序列库，将其目标误差转化为策略对数几率建议，且仅当无标度预测可靠性证书、策略边界门和argmax一致性门三者均通过时，才会采纳该建议；否则将完全回退至自身基于模型的基础控制器。在CartPole-v1和离散化Pendulum-v1环境中，针对物理参数偏移，本文在9种受控基准（包括Transformer、循环、拆分隐变量、图、因果归纳模型及3种近期基于模型的推理模块）上进行评估，采用共享的PPO训练器、相同的交互预算及10个保留种子（共200次运行）。CausalNav取得了最佳平均排名（10个基准中为1.25）。诊断结果比排名更具信息量：学习到的图恢复结构的表现远高于随机水平（CartPole的F1值为0.59±0.09），但每个种子的结构保真度与对应种子的控制效益无相关性（相关系数r=-0.15，p=0.67），且在10个Pendulum种子上，证书均会弃权（不执行），此时强制规划器执行会降低回报。在本文设定中，模型保真度无法预测下游控制效用；经认证的弃权（不执行）而非更好的预测，才是世界模型可安全部署的关键。
英文摘要：
A world model is only useful for physical AI if it changes what the agent does, and only safe if it declines to do so when it is wrong. We study both halves of that requirement with CausalNav, a controller built around a signed, action-conditioned transition graph over identified state coordinates. At deployment CausalNav simulates a small library of intervention sequences, converts their objective error into policy-logit advice, and admits that advice only when a scale-free predictive-reliability certificate, a policy-margin gate, and an argmax-agreement gate all pass; otherwise it falls back exactly to its own model-based base controller. We evaluate against nine controlled baselines (transformer, recurrent, split-latent, graph, causal-induction, and three recent model-based reasoning modules) on CartPole-v1 and discretized Pendulum-v1 with physical-parameter shifts, under one shared PPO trainer, one interaction budget, and ten held-out seeds (200 runs). CausalNav attains the best average rank (1.25 of ten). The diagnostic result is more informative than the ranking: the learned graph recovers structure well above chance (CartPole F1 = 0.59 +/- 0.09), yet per-seed structural fidelity is uncorrelated with per-seed control benefit (r = -0.15, p = 0.67), and the certificate abstains on 10/10 Pendulum seeds, where forcing the planner on costs return. Model fidelity did not predict downstream control utility in our setting; certified abstention, not better prediction, is what made the world model safe to deploy.
