arXiv：cs.LG· Amit Thakur, Mukesh Singhal·· 3 小时前AI 评分33
多智能体 Transformer 策略的排列鲁棒性不够：动作坍缩问题
Permutation Robustness Is Not Enough: Action Collapse in Multi-Agent Transformer Policies
AI 导读
多智能体 Transformer 策略仅靠低排列误差可能产生误导，因为所有智能体选择相同动作也会显得鲁棒。研究提出结合排列一致性指标与动作坍缩诊断（动作多样性、同动作比例、最大动作频率）来评估策略。PPO-ID 基线行为未坍缩但对顺序敏感，强等变性正则化仍可导致同质行为；弱等变性惩罚在 N=3 时提升鲁棒性并保留动作多样性，N=4 时则需更小的正则化权重。
正文
Abstract:Transformer policies are attractive for multi-agent robot learning because self-attention can model interactions among agents. However, multi-agent teams are unordered, while transformers typically process agents as ordered token sequences. We study how this mismatch affects cooperative navigation policies under agent-order permutations. Our results show that low permutation error alone can be misleading: policies may appear robust simply because all agents choose the same action. We therefore evaluate policies using both permutation-consistency metrics and action-collapse diagnostics, including action diversity, same-action fraction, and maximum action frequency. A PPO-ID baseline yields non-collapsed behavior but remains order-sensitive, while strong equivariance regularization can still induce homogeneous behavior. A weak equivariance penalty improves the robustness while preserving more diverse actions for teams with \(N=3\) agents, whereas teams with \(N=4\) agents require substantially smaller regularization weights. These findings suggest that multi-agent transformer policies should be evaluated not only by return and permutation robustness, but also by whether they maintain non-collapsed, differentiated multi-agent behavior.
| Subjects: | Robotics (cs.RO); Machine Learning (cs.LG); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2610.02848 [cs.RO] |
| (or arXiv:2610.02848v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02848 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Amit Thakur [view email]
[v1]
Fri, 2 Oct 2026 05:35:53 UTC (44 KB)
来源：arXiv：cs.LG · arxiv.org