AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

CoRA: A Collaborative Robust Architecture with Hybrid Fusion for Efficient Perception

Gong Chen, Chaokun Zhang, Pengcheng Lv, Xiaohui Xie

AAAI 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Collaborative multi-agent perception · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add PAC object-level correction branch (ablation)CoRA (CIT+LC+L_align) · DAIR-V2X, pose error 0.4/0.4AP@0.7: 0.5371 -> 0.5862Improved
add PAC object-level correction branch (ablation)CoRA (CIT+LC+L_align) · DAIR-V2X, no pose errorAP@0.7: 0.6359 -> 0.6361No clear effect
add LC module on top of CIT (ablation)CoRA · DAIR-V2X, no pose errorAP@0.7: 0.6036 -> 0.6288Improved
add alignment (distillation) loss L_align to LC (ablation)CoRA (CIT+LC) · DAIR-V2X, no pose errorAP@0.7: 0.6288 -> 0.6359Improved
add LC and L_align on top of CIT under noise (ablation)CoRA · DAIR-V2X, pose error 0.4/0.4AP@0.7: 0.5184 -> 0.5371Improved
use winner-take-all CIT instead of MaxOut (ablation)CoRA · DAIR-V2X, no pose errorAP@0.7: 0.5610 -> 0.6359Improved
use Top-1 instead of Top-2 selection in CIT (ablation)CoRA · OPV2V, no pose errorAP@0.7: 0.8652 -> 0.8661No clear effect
use Top-1 instead of Top-2 selection in CIT (ablation)CoRA · DAIR-V2X, no pose errorAP@0.7: 0.6211 -> 0.6359Improved
CoRA hybrid design vs intermediate-only baselines under pose errorcollaborative perception · OPV2V, pose error 0.6/0.6AP@0.7: 0.6544 vs 0.4849 (CoAlign)Improved
add object-level correction branch when pose error is minimalCoRA · OPV2V, low pose error 0.2/0.2AP@0.5: slight perturbation versus intermediate-only branchMixed
use CoRA under communication latencyCoRA vs ERMVP · OPV2V, pose error 0.6/0.6, 300ms delayAP@0.7: 0.1705 -> 0.3651Improved
add CIT and LC feature-level modules to single-agent baseline (ablation)single-agent detector (ResNet backbone) · DAIR-V2X, no pose errorAP@0.7: 0.4457 -> 0.6359Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy