CauseCollab Uses Causal Metric Learning for Heterogeneous Collaborative Perception

CauseCollab addresses semantic inconsistency in multi-agent sensor sharing by disentangling semantic factors from modality-specific statistical confounders. It uses causal metric learning to map heterogeneous features into a shared protocol space, reducing error accumulation during modality conversion.