Researchers propose CoAnchor, an anchor-centric framework for collaborative perception in autonomous vehicles. This approach addresses the challenges of spatial and temporal misalignment by building sparse object-level anchors as a shared interface for pose correction. The method tightly connects spatial refinement, temporal propagation, and current-time verification within one unified loop, while keeping the overall process lightweight. Experiments on simulated and real-world datasets show that CoAnchor remains competitive under clean settings and improves robustness under joint delay and pose perturbations. --- Why it matters: This matters to AI researchers because it tackles a critical challenge in autonomous driving: handling asynchronous collaborative perception with spatial and temporal misalignment. The proposed framework has the potential to improve the robustness of autonomous vehicles in real-world scenarios. Source: https://arxiv.org/abs/2608.21055