Zihan Wang

Robotics Institute, Carnegie Mellon University

I'm a Master's student in Computer Vision (MSCV) at the Robotics Institute, School of Computer Science, Carnegie Mellon University, advised by Prof. Deva Ramanan. My research focuses on 4D reconstruction and its applications to robotics.

Prior to CMU, I obtained my B.Eng. with Honours in Electrical and Electronic Engineering from the University of Edinburgh, where I worked on counterfactual estimation with the Causality in Healthcare AI Hub. During my undergraduate years, I also spent time at AIRLab working with Prof. Sebastian Scherer and Prof. Chen Wang.

I build machine learning, computer vision, and robotics systems that close the loop between perception and action. If you're exploring robotics projects that need strong vision capabilities, feel free to drop me a line—I'm always open to collaboration.

Email  /  Google Scholar  /  GitHub  /  LinkedIn  /  X

We live in a dynamic 4D world, constantly seeing, understanding, and interacting with objects and environments. My goal is to bridge the gap between reconstructed virtual worlds and practical robotic applications. I develop methods that help robots perceive, understand, and interact with objects and environments from sparse observations.

A real-to-sim framework that turns monocular RGB video into whole-body control across diverse, complex terrains.

The first approach that enables free-view synthesis for 4D dynamic scene reconstruction under sparse-view capture.

Class-agnostic relations for few-shot detection without fine-tuning, enabling fast and efficient field deployment.

A general framework that learns whole-body humanoid-object interaction skills directly from monocular RGB videos.

Characterizes noise in large-scale pre-training data and proposes strategies to mitigate its impact on downstream tasks.

Cross-sectional counterfactual estimation conditioned on disentangled identity-preserving features.

A simple, effective, fine-tuning-free strategy to pick the optimal diffusion depth for each sample.

Bridges classic robotics and modern learning with a physics-aware optimization toolkit.

Clusters motion data, trains latent variable models, and deploys a style-aware high-level controller for multi-task behaviors.

Contributed physics-aware constrained optimization modules for trajectory planning and motion control.

Developed diffusion autoencoder components for medical imaging, with a focus on MRI generation and reconstruction.

• Edinburgh Award (Research Experience), 2022
• Edinburgh Scholarship (£4,000), 2022
• Edinburgh Scholarship (£4,000), 2021
• QIANJIANG ELECTRIC Scholarship (£3,000), 2020

Academic Reviewer

• International Conference on Learning Representations (ICLR)
• European Conference on Computer Vision (ECCV)
• IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
• IEEE Transactions on Image Processing (TIP)
• IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)

• Email: zihanwa3@cs.cmu.edu
• Phone: +1-412-589-8046
• Address: EDSH, Carnegie Mellon University, Pittsburgh, PA 15213, USA

I am confident that my "boundaryless", "website-stalking" friends have contributed to at least 74.47% of the traffic here.

Template adapted from Qitao Zhao (credit to Yijia Weng & Jon Barron).
Last updated: February 2025