Xiaopeng Wu 吴小鹏

I am a graduate researcher in Computer Science at Binghamton University, working in the Autonomous Intelligent Robotics (AIR) Lab under the supervision of Prof. Shiqi Zhang. My research focuses on robot learning for humanoid locomotion and manipulation.

I build systems that connect human demonstrations with learned robot behavior. My work spans visual teleoperation, demonstration data, and imitation learning, with the long-term goal of general-purpose humanoid loco-manipulation.

Before graduate research, I spent eight years building production software and large-scale data platforms, including four years at PayPal. I bring software engineering, distributed systems, and production reliability experience to robotics research.

Current Research

Learning Humanoid Control from Visual Teleoperation

G1 Teleop Data Platform · 2026–Present

I develop a platform for collecting human demonstrations and training humanoid task policies. Current demonstrations include whole-body teleoperation, autonomous target interaction, and human-operated bimanual box relocation in simulation.

Bimanual box relocation Through visual teleoperation, an operator guides the robot to lift, carry, and place a box.
Whole-body teleoperation Human motion drives the simulated robot in real time.
Autonomous target interaction A learned policy approaches a target, touches it, and returns to standing.

News

Experience

Graduate Student Researcher

Autonomous Intelligent Robotics (AIR) Lab, Binghamton University · Advisor: Prof. Shiqi Zhang

Research robot learning for humanoid locomotion and manipulation, connecting visual teleoperation, demonstration data, and imitation learning to help robots acquire new skills from human demonstrations.

Software Engineer

PayPal, Shanghai

Built a data platform automating over 200 TB of daily processing for data engineering teams, integrating monitoring and resource optimization to improve efficiency and reliability. Led a five-person team through a 2 PB enterprise data migration to the cloud.

Software Engineer

PingAn Medical & Healthcare, Shanghai

Developed a data integration and scheduling platform that replaced more than 500 standalone scripts and automated large-scale data workflows. Built data synchronization pipelines processing over 1 TB daily to keep downstream systems up to date.