Carter Ung

Engineering at Microsoft · Embodied AI & Robotics at University of Washington

I am a predoctoral researcher at the University of Washington, working in the Robotics and State Estimation Lab and Personal Robotics Lab under Dieter Fox (Sr. Director, Ai2) and Siddhartha Srinivasa (Partner, Madrona Ventures). My research centers on robot manipulation: building action policies and learning systems that leverage rich, diverse data representations for embodied reasoning, task generalization, and long-horizon planning. I am supported by the NSF CISE Graduate Fellowship.

I also work as a software engineer at Microsoft, building Copilot into OneNote. Previously, I studied Computer Science and Biomedical Engineering at the University of Houston, where I spent over two years with Dr. Shishir Shah on pose-invariant face recognition (VISAPP 2025).

I’ve come across many kind people in my curious journey toward academia and industry. I am always open to chat and talk about perspectives in research, career, and life aspirations. Reach me at carterung [at] gmail [dot] com.

News

January 2026: RoboEval accepted to IEEE International Conference on Robotics and Automation (ICRA) 2026!

August 2025: Selected as a 2025 National Science Foundation Computer and Information Science and Engineering Graduate Fellow!

Publications

	RoboPlayground: Democratizing Robotic Evaluation through Structured Physical Domains Yi Ru Wang^, Carter Ung^, Evan Gubarev, Christopher Tan, Siddhartha Srinivasa^†, Dieter Fox^† ^*Equal contribution ^†Equal advising Coming Soon
	RoboEval: Where Robotic Manipulation Meets Structured and Scalable Evaluation Yi Ru Wang, Carter Ung, Christopher Tan, Grant Tannert, Jiafei Duan, Josephine Li, Amy Le, Rishabh Oswal, Markus Grotz, Wilbert Pumacay, Yuquan Deng, Ranjay Krishna, Dieter Fox^†, Siddhartha Srinivasa^† ^†Equal advising IEEE Conference on Robotics and Automation 2026 (ICRA), Eval&Deploy Workshop Paper at CoRL (Oral Spotlight), Bimanual Manipulation: Advancing Human-Humanoid Interaction and Collaboration Workshop at IEEE International Conference on Intelligent Robots and Systems (IROS 2025) Website PDF Code RoboEval introduces a unified evaluation suite covering bimanual manipulation tasks with fine-grained metrics so researchers can compare models beyond binary success/failure.
	Minimizing Number of Poses for Pose-Invariant Face Recognition Carter Ung, Pranav Mantini, Shishir Shah International Conference on Computer Vision Theory and Applications (VISAPP) 2025 PDF We study how many viewpoint images are truly needed for robust, pose-invariant face recognition and show that a compact pose set dramatically reduces data-collection overhead.
	Dual-VXM: A Prior-Aware Dual-Encoder UNet for Watch-and-Wait Monitoring with Low-Field MRI Mohammad Javadi, Panagiotis Tsiamyrtzis, Carter Ung, Ernst Leiss, Andrew Webb, Nikolaos Tsekos Under Review We present a novel dual-encoder UNet architecture that leverages prior high-field MRI scans to enhance low-field MRI image quality for improved monitoring of Low-Grade Gliomas (LGG).
	Minimizing the Number of Poses for Pose-Invariant Face Recognition Carter Ung Undergraduate Dissertation, University of Houston 2024 PDF My honors thesis extends the VISAPP study and details the full experimental pipeline for low-redundancy face-data acquisition.