Temporal Residual Bottleneck for Robust Asynchronous Collaborative Perception

2026-10-07

AI Summary

This study addresses the performance degradation in cooperative perception caused by communication delays and packet loss, which result in stale and incomplete feature representations. To mitigate this issue, asynchronous cooperative perception is reformulated as a temporal residual prediction task, and a Temporal Residual Bottleneck architecture is proposed. Specifically, pose-aligned features serve as conservative anchors to avoid directly transmitting unreliable features, while xLSTM networks extract historical residual evidence. The predicted residuals are subsequently refined via gated regularization before fusion, achieving an effective balance between static structural preservation and dynamic compensation. Extensive experiments on the DAIR-V2X and OPV2V datasets demonstrate that the proposed method achieves superior performance under severe fixed or irregular delays and packet loss, significantly enhancing perceptual robustness in communication-degraded scenarios.

Abstract

Collaborative perception extends the sensing range of autonomous vehicles, but its performance degrades when shared features arrive stale or incomplete. Most latency-robust methods compensate delayed collaborator features through flow-guided alignment or direct feature transport. In this work, we formulate asynchronous collaborative perception as temporal residual prediction. Our Temporal Residual Bottleneck keeps a deterministic pose-warped collaborator feature as a conservative anchor and uses a $Δt$-conditioned xLSTM to extract residual temporal evidence from the available history. A detector-facing residual bottleneck then applies only gated, regularized corrections before ego-side fusion, reducing the risk of overwriting reliable static structure when temporal correspondence is uncertain. Experiments on DAIR-V2X and OPV2V show that our method is especially effective under severe fixed/irregular delays and packet drops. On DAIR-V2X, the reported checkpoint trades a small amount of synchronized peak accuracy for better robustness under stronger communication degradation. Controlled diagnostics further indicate that direct feature transport has oracle headroom but can become unreliable when deployed without accurate correspondence. These results support temporal residual fusion as a practical alternative for asynchronous and incomplete collaborative perception. Code will be publicly released at https://url.fzi.de/8dk38.

Problem

• collaborative perception
• asynchronous communication
• latency robustness
• packet loss
• autonomous driving

Innovation

• Temporal Residual Bottleneck
• Asynchronous Collaborative Perception
• xLSTM
• Latency Robustness
• Residual Fusion