PIPO: Policy Optimization with Permutation-Invariant Constraint for Distributed Multi-Robot Navigation | IEEE Conference Publication | IEEE Xplore