Yangcheng Yu

Curriculum Vitae

Beijing, China
2023.09 – 2027.07 (Expected)

Tsinghua University
Bachelor of Engineering in Electronic Engineering
GPA: 3.75/4.0 (Rank: 46/266, Top 20%)

Singapore
2026.01 – Present

Beijing, China
2025.06 – 2026.01

Research Intern, Manifold AI
- Developed a unified Reasoner-Dreamer-Actor architecture for the WMA 3.0 project.
- Designed the Action VAE model to bridge the gap between diverse robot embodiments and human demonstrations.
- Processed over 10 million action trajectories across 5+ heterogeneous robotic platforms.
- Proposed MoWM, achieving state-of-the-art results on the CALVIN benchmark with a success rate improvement of +5.7% to +13.4%.

Beijing, China
2025.06 – 2025.09

Research Assistant, Institute for AI Industry Research (AIR), Tsinghua University
- Co-designed the PAM engine, disentangling pose, appearance, and motion for photorealistic HOI video generation.
- Implemented multi-conditional geometric control mechanisms (depth maps, segmentation masks, hand keypoints), improving geometric consistency by up to 51.6%.
- Curated the Impromptu VLA dataset (80K+ clips) for autonomous driving in long-tail urban scenarios.

Beijing, China
2024.11 – 2025.05

Research Assistant, Future Intelligence Lab (FIBLab), Tsinghua University
- Built a full-stack robotic system based on ROS2 and Xiaomi Cyberdog for vision-language navigation (VLN).
- Developed remote teleoperation and communication pipelines for real-world data collection in open urban environments.

MoWM: Mixture-of-World-Models for Embodied Planning via Latent-to-Pixel Feature Modulation
Yu Shang*, Yangcheng Yu*, et al.
ICML 2026 (Under Submission)
PAM: A Pose–Appearance–Motion Engine for Sim-to-Real HOI Video Generation
Mingju Gao, Kaisen Yang, Huan-ang Gao, Bohan Li, Ao Ding, Wenyi Li, Yangcheng Yu, et al.
CVPR 2026
Impromptu VLA: Open Weights and Open Data for Driving Vision-Language-Action Models
Haohan Chi, Huan-ang Gao, Ziming Liu, Jianing Liu, Chenyu Liu, Jinwei Li, Kaisen Yang, Yangcheng Yu, et al.
NeurIPS 2025 (Datasets and Benchmarks Track)
Learning to Navigate in Open Urban Environments Using a Simple Sim2Real Strategy
Lixuan He, Haoyu Dong, Yangcheng Yu, et al.
ICLR 2025 Workshop on EmbodiedAI

2024, 2025

Innovation Excellence Scholarship (National Endeavor Scholarship), Tsinghua University

2024

2022