Qixin Zhang receives Best Paper Runner-Up Award at MEAS Workshop

Department of Computer Science PhD candidate Qixin Zhang received the Best Paper Runner-Up Award at the Multi-Agent Embodied Intelligent Systems (MEAS) Workshop at the 2026 Computer Vision and Pattern Recognition Conference (CVPR). 

“It's rewarding to see our work recognized by the community, especially because it addresses a practical challenge in real-world autonomous systems,” Zhang said. “This recognition motivates me to continue pursuing research that bridges theory and deployment.”

Zhang's paper, titled “Multi-Agent Video Prediction: Self-Correcting Conditional Frames for Dynamic Scene Forecasting,” looks at the issue of video delays in remote driving due to uplink constraints. The paper outlines a multi-agent system solution to the problem. One agent predicts upcoming video frames, another detects new objects in the scene, and a third corrects the prediction system using sparse mask guidance. 

Tested under real-world 5G network traces, the approach significantly improved the system's ability to handle unexpected changes while preserving visual quality and real-time performance. 

“I hope it can help make remote assistance for autonomous vehicles more efficient and scalable in real-world deployments,” Zhang said. “I enjoy working on open-ended problems and building solutions that can have real-world impact.”

To students considering research, Zhang encourages perseverance.

“Find a problem that genuinely interests you and don't get discouraged when things don't work at first,” Zhang said. “Research often involves many failed attempts before finding something that works.”

In receiving Best Paper Runner-Up, Zhang expresses thanks to her advisor, Professor Zhi-Li Zhang, for his guidance and support. She is also grateful to Postdoctoral Associate Ajay Kumar Gurumadaiah for his leadership and contributions to the project and insightful discussions. 

Learn more about Zhang's research at her Google Scholar profile.