When the Score Becomes the Target: Rethinking Metric Validity in Autonomous Driving
Do benchmark score gains still reflect real driving improvements once the score becomes the optimization target? Controlled experiments say: not necessarily.
Publications
Research papers on cooperative autonomous driving, 3D vision, and scene understanding.
LANTERN: A Closed-Loop Benchmark for VLM-Based Cooperative Driving with Temporally Grounded Warnings
A closed-loop benchmark evaluating how vision-language models respond to temporally grounded safety warnings in cooperative driving.
F3DGS: Federated 3D Gaussian Splatting for Decentralized Multi-Agent World Modeling
A federated framework that lets multiple agents collaboratively reconstruct 3D scenes with Gaussian splatting without centralizing raw data.
Mind the Hitch: Dynamic Calibration and Articulated Perception for Autonomous Trucks
dCAP: a vision-based framework that continuously estimates the 6-DoF relative pose between tractor and trailer cameras, plus the STT4AT benchmark for articulated trucks.
M³CAD: Towards Generic Cooperative Autonomous Driving Benchmark
A generic benchmark for cooperative autonomous driving supporting multi-vehicle, multi-modal, and multi-task evaluation.
From Features to Reference Points: Lightweight and Adaptive Fusion for Cooperative Autonomous Driving
A lightweight and adaptive reference-point-based fusion approach for cooperative autonomous driving.
Investigating Crowdsourced Neural Radiance Maps for Autonomous Vehicles
Investigating the use of crowdsourced neural radiance fields (NeRF) as maps for autonomous vehicles.
A Global Multi-Temporal Dataset with STGAN Baseline for Cloud and Cloud Shadow Removal
A global multi-temporal satellite dataset with an STGAN baseline for cloud and cloud shadow removal.
Process Monitoring and End-Point Identification of Unattended Liquid Phase Separation Based on Edge Detection
Edge-detection-based process monitoring and end-point identification for unattended liquid phase separation.