Papers

Permutation Robustness Is Not Enough: Action Collapse in Multi-Agent Transformer Policies

This preprint examines transformer policies for cooperative multi-robot navigation and shows that measuring permutation equivariance alone is misleading: low error can occur because all agents output the same action, not because the policy is robust. In a PettingZoo/MPE cooperative navigation task, a PPO baseline with learned agent ID/position embeddings is diverse but order-sensitive, while equivariance regularization reduces order error but can collapse actions, especially when moving from 3 to 4 agents. The work argues for reporting task return, permutation consistency, and action diversity together.