Evaluation of an Uncertainty-Aware Late Fusion Algorithm for Multi-Source Bird's Eye View Detections Under Controlled Noise

2025/07/04 by Fadili, Maryem, Lecrosnier, Louis, Pechberti, Steve +1
#FOS: Computer and information sciences #FOS: Electrical engineering #Robotics (cs.RO) #Signal Processing (eess.SP) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2507.03381

Abstract

Reliable multi-source fusion is crucial for robust perception in autonomous systems. However, evaluating fusion performance independently of detection errors remains challenging. This work introduces a systematic evaluation framework that injects controlled noise into ground-truth bounding boxes to isolate the fusion process. We then propose Unified Kalman Fusion (UniKF), a late-fusion algorithm based on Kalman filtering to merge Bird's Eye View (BEV) detections while handling synchronization issues. Experiments show that UniKF outperforms baseline methods across various noise levels, achieving up to 3x lower object's positioning and orientation errors and 2x lower dimension estimation errors, while maintaining nearperfect precision and recall between 99.5% and 100%.

Citations

• CooPre: Cooperative Pretraining for V2X Cooperative Perception
• V2X-DGW: Domain Generalization for Multi-agent Perception under Adverse Weather Conditions
• TUMTraf V2X Cooperative Perception Dataset
• CoBEVFusion: Cooperative Perception with LiDAR-Camera Bird's-Eye View Fusion
• Probabilistic 3D Multi-Object Cooperative Tracking for Autonomous Driving via Differentiable Multi-Sensor Kalman Filter
• InfraDet3D: Multi-Modal 3D Object Detection based on Roadside Infrastructure Camera and LiDAR Sensors
• Collaborative Multi-Object Tracking with Conformal Uncertainty Propagation
• Uncertainty Quantification of Collaborative Detection for Self-Driving
• Delving into the Devils of Bird's-eye-view Perception: A Review, Evaluation and Recipe
• CO³: Cooperative Unsupervised 3D Representation Learning for Autonomous Driving
• DAIR-V2X: A Large-Scale Dataset for Vehicle-Infrastructure Cooperative 3D Object Detection
• Model-Agnostic Multi-Agent Perception Framework
• V2X-Sim: Multi-Agent Collaborative Perception Dataset and Benchmark for Autonomous Driving
• Multi-modal Sensor Fusion for Auto Driving Perception: A Survey
• A Survey of Uncertainty in Deep Neural Networks
• A Competitive Method to VIPriors Object Detection Challenge
• CLOCs: Camera-LiDAR Object Candidates Fusion for 3D Object Detection
• AB3DMOT: A Baseline for 3D Multi-Object Tracking and New Evaluation Metrics
• When2com: Multi-Agent Perception via Communication Graph Grouping
• Derivation of a Constant Velocity Motion Model for Visual Tracking
• nuScenes: A multimodal dataset for autonomous driving
• Generalized Intersection over Union: A Metric and A Loss for Bounding\n Box Regression
• Joint 3D Proposal Generation and Object Detection from View Aggregation
• Soft-NMS -- Improving Object Detection With One Line of Code
• Multi-View 3D Object Detection Network for Autonomous Driving
• Vision meets robotics: The KITTI dataset

Related

• Delving into the Devils of Bird's-eye-view Perception: A Review, Evaluation and Recipe
• UniFusion: Unified Multi-view Fusion Transformer for Spatial-Temporal Representation in Bird's-Eye-View
• A Black-Box Evaluation Framework for Semantic Robustness in Bird's Eye View Detection
• OnlineBEV: Recurrent Temporal Fusion in Bird's Eye View Representations for Multi-Camera 3D Perception
• FusionFormer: A Multi-sensory Fusion in Bird's-Eye-View and Temporal Consistent Transformer for 3D Object Detection