# HyDRA: Hybrid Domain-Aware Robust Architecture for Heterogeneous Collaborative Perception Source: [HyDRA: Hybrid Domain-Aware Robust Architecture for Heterogeneous Collaborative Perception]() Collection index: [llms.txt]() Data updated: 2026-03-27T05:00:04.194Z Authors: Minwoo Song, Minhee Kang, Heejin Ahn Published: 2026-03-25T06:15:25.000Z ## Abstract In collaborative perception, an agent's performance can be degraded by heterogeneity arising from differences in model architecture or training data distributions. To address this challenge, we propose HyDRA (Hybrid Domain-Aware Robust Architecture), a unified pipeline that integrates intermediate and late fusion within a domain-aware framework. We introduce a lightweight domain classifier that dynamically identifies heterogeneous agents and assigns them to the late-fusion branch. Furthermore, we propose anchor-guided pose graph optimization to mitigate localization errors inherent in late fusion, leveraging reliable detections from intermediate fusion as fixed spatial anchors. Extensive experiments demonstrate that, despite requiring no additional training, HyDRA achieves performance comparable to state-of-the-art heterogeneity-aware CP methods. Importantly, this performance is maintained as the number of collaborating agents increases, enabling zero-cost scaling without retraining. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [3D Object Detection on V2X-Real Gaussian Pose Noise σ = 0.4 (test)](): 3D Object Detection; V2X-Real Gaussian Pose Noise σ = 0.4 (test); AP@0.3 (Vehicle): 84.27. - [3D Object Detection on V2X-Real Gaussian Pose Noise σ = 0.4](): 3D Object Detection; V2X-Real Gaussian Pose Noise σ = 0.4; AP@0.3 (vehicle): 84.27. - [3D Object Detection on V2X-Real Robustness Analysis](): 3D Object Detection; V2X-Real Robustness Analysis; AP@0.3 (Vehicle): 0.8427. - [3D Object Detection on V2X-Real Ideal Setting (test)](): 3D Object Detection; V2X-Real Ideal Setting (test); AP@0.3 (Vehicle): 86.84. - [3D Object Detection on V2X-Real Ideal Setting](): 3D Object Detection; V2X-Real Ideal Setting; AP@0.3 (Vehicle): 86.84. - [3D Object Detection on V2X-Real Ideal Setting (No Noise) (test)](): 3D Object Detection; V2X-Real Ideal Setting (No Noise) (test); AP@0.3 (Vehicle): 86.84. ## Original sources and implementation [Original paper]() [PDF]()